FG7142 ATTENUATES EXPRESSION OF OVEREXPECTATION IN
PAVLOVIAN FEAR CONDITIONING
Joshua Benjamin Bernard Garfield
December 2008
This thesis has been submitted in fulfilment of the requirements for the degree of
Doctor of Philosophy
School of Psychology
University of New South Wales
Abstract
The experiments reported in this thesis studied the mechanisms of expression of overexpectation of conditioned fear, as measured by freezing. In Stage I, rats were conditioned to fear a tone and a flashing light conditioned stimulus (CS) through pairings with a 0.5 mA, 1 s shock. In Stage II, overexpectation was trained by the reinforcement of a compound of these CSs with a shock of the same magnitude. Two compound – shock pairings produced an overexpectation effect, as measured by freezing to presentations of the tone alone, while further Stage II training caused over- training of overexpectation. Expression of the overexpectation effect produced by two compound – shock pairings could be prevented by pre-test injection of the benzodiazepine partial inverse agonist FG7142. This effect was dose-dependent and not due to state-dependent memory. Control experiments suggested that it was also not due to any general effect of FG7142 on the Pavlovian freezing response. Freezing to a tone that had been conditioned, but not subjected to any decremental training procedures, was unaffected by administration of FG7142 before either the conditioning or test session. FG7142 also did not affect freezing to a tone that had been subjected to an associative blocking procedure. The hypothesis that overexpectation of conditioned fear may be context-dependent was also tested. However, renewal was not observed. Rats that received Stage II training in a context distinct from the Stage I training context showed equivalent expression of overexpectation regardless of whether testing was conducted in the Stage I or Stage II training context. These results are consistent with the hypothesis that overexpectation, like extinction, leads to the imposition of a GABAA receptor-mediated mask on the fear CR. Moreover, they suggest that this masking of fear is the specific consequence of negative predictive error.
i Table of Contents
Abstract i
Certificate of Originality vi
Acknowledgements vii
Manuscript and Conference Presentations ix
Care and Use of Animals x
List of Tables xi
List of Figures xii
Abbreviations xiv
Introduction 1
Chapter 1. Behavioural characteristics of the learning and loss of fear 4
1. Acquisition of Pavlovian fear conditioning 5
1.1 The role of error-correction in the acquisition of conditioned fear 8
2. Extinction of Pavlovian fear conditioning 13
2.1 The contribution of error-correction to extinction 14
2.2 Reinstatement 17
2.3 Renewal 19
2.4 Spontaneous recovery 22
2.5 Theories of extinction 23
3. Overexpectation 31
4. Chapter 1 conclusion 44
ii Chapter 2. The neurobiology of the acquisition, expression, and loss of 47 conditioned fear
1. The role of the amygdala in the acquisition of conditioned fear 49
1.1. Theories regarding the role of the amygdala in storage of fear 53
memory
1.2. The neurochemistry of fear conditioning 55
2. The neuroanatomy of the expression of conditioned fear 65
3. Opioid receptor activation modulates detection of prediction error in 69
both acquisition and extinction of conditioned fear
4. The role of the amygdala in extinction 74
5. The prefrontal cortex in extinction 84
5.1. Plasticity of IL neural activity is associated with extinction 88
5.2. Role of NMDA receptors and intracellular signalling in mPFC in 90
extinction
6. Inhibition of amygdala neurons by IL activity as a possible substrate 92
for extinction
7. The role of the hippocampus in contextual modulation of extinction 94
8. Summary of models of extinction and conclusion 96
Chapter 3. Experimental reports 102
Experiment 1 102
Experiment 2 111
Experiment 3 118
Experiment 4 123
Experiment 5 131
Experiments 6a and 6b 135
iii Experiment 7 141
Experiments 8a and 8b 145
Experiment 9 150
Experiment 10 157
Chapter 4. General Discussion 162
1. Summary of empirical results and their theoretical implications 162
2. The over-training of overexpectation 166
3. Theoretical implications of the effect of FG7142 on expression of 172
overexpectation
3.1 A role for a comparator process? 173
3.2. How is fear masked after overexpectation? 175
4. What are the neuroanatomical substrates for the expression of 182
overexpectation?
5. The relationship between the pharmacology of FG7142 and the 184
disinhibition of fear
6. Conclusion 189
References 191
Appendix 1: Experiment 1 219
Appendix 2: Experiment 2 241
Appendix 3: Experiment 3 254
Appendix 4: Experiment 4 270
Appendix 5: Experiment 5 287
Appendix 6: Experiment 6a 294
Appendix 7: Experiment 6b 300
Appendix 8: Experiment 7 311
iv Appendix 9: Experiment 8a 321
Appendix 10: Experiment 8b 328
Appendix 11: Experiment 9 334
Appendix 12: Experiment 10 375
v Certificate of Originality
ORIGINALITY STATEMENT ‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’
Signed ……………………………………………......
Date ……………………………………………......
vi Acknowledgements
First and foremost, I must acknowledge the indispensable contribution to this project, and my education in learning theory and neuroscience in general, by my supervisor, Dr. Gavan Patrick McNally. Thank you very much for your careful attention to my progress, my written work and presentations, and my experimental designs and results at every stage of my candidature. Thank you for your help, so many times, in charting the way forward, and your encouragement and reassurance when things didn’t seem to be going so well. It would be difficult, if not impossible, to understate your contribution to all that I have learned about this field in the past four years, and for stimulating my interest in it.
Of course, I would never have come anywhere near the stage of doing a PhD if it were not for the love and support of my parents, Jay Lazar Garfield and Blaine Garson.
Dad, since I was a toddler, you have actively stimulated my interest in science.
Everything you have done, from teaching me about the different varieties of moss in the forest in Belchertown, to explaining the different types of infinities, to buying me books on astronomy, has contributed to the realisation of this thesis. Mom, your wisdom, encouragement, and belief in me were indispensable to getting me here.
Thanks also to my brother, Abra James Coffin Garfield for the varied ways he supported me while I’ve lived in Sydney. Thanks for your supportive interest in my topic of study, which has helped maintain my own interest in it and shown me novel ways to think about it. And thank you most of all for getting me back into Ultimate, without which my experience of Sydney would have been considerably less colourful.
vii To my partner, Troy-John William Emery, thank you for your love and support through all the trials and tribulations that have been part of my PhD studies. You endured 1 ½ years of a long-distance relationship, plus all my varied moods and eccentricities for much longer, and I’m so glad I can now share this ‘victory’ with you.
Your affection and sense of humour are a haven for me amidst all the challenges I’ve faced.
Thanks also to my comrades for their understanding and support. Your efforts inspire me with the hope that we can create a world where science will function to benefit all of humanity, rather than being distorted around the profit-motives of the corporate elite.
Finally, thanks to all my office- and lab-mates, and various denizens of Level 6 for the innumerable ways in which they have helped, supported, and taught me. I could probably thank nearly every person who’s been involved in the McNally, Richardson, and Westbrook labs in the past 4 years, along with the rat attendants who have worked here in that time, and others in the School of Psychology as well. However, I’d particularly like to thank Melissa Wood, Kate Blatchford, Laura Bradfield, Sindy Cole,
Julia Langton, Glynis Bailey, Lucy Choi, Marianne Weber, Adam Hamlin, Kelly
Clemens, Christina Perry, Genevrie Hart, Jee Hyun Kim, and, finally, my co-supervisor
Fred Westbrook.
Thanks also to Justin Harris for his useful suggestions regarding possible interpretations of some of the data presented in this thesis.
viii Manuscript
Garfield, J. B. B. & McNally, G. P. (In Press). The effects of FG7142 on overexpectation of Pavlovian fear conditioning. Behavioral Neuroscience.
Conference Presentations
Garfield, J. B. B., & McNally, G. P. (2007). The role of GABAA receptor function in the expression of over-expectation. International Brain Research Organization World
Congress of Neuroscience, Melbourne.
Garfield, J. B. B. & McNally, G. P. (2007). FG-7142 prevents expression of over- expectation in Pavlovian fear conditioning. Neuroscience 2007, San Diego.
Garfield, J. B. B. & McNally, G. P. (2008). FG7142, a benzodiazepine receptor partial inverse agonist, prevents expression of over-expectation in Pavlovian fear conditioning.
28th Annual Meeting of the Australian Neuroscience Society, Hobart.
ix Care and Use of Animals
The experiments presented in this thesis conformed to the guidelines on the ethical use of animals maintained by the Australian Code of Practice for the Care and Use of
Animals for Scientific Purposes (7th Edition), and all procedures were approved by the
Animal Care and Ethics Committee at the University of New South Wales. All efforts were made to minimise both suffering and the number of animals used.
x List of Tables
Table Title Page
1 Design of Experiment 1. 107
2 Design of Experiment 2. 115
3 Design of Experiment 3. 121
4 Design of Experiment 4. 127
5 Design of Experiment 5. 134
6 Design of Experiment 6a. 138
7 Design of Experiment 6b. 140
8 Design of Experiment 7 144
9 Design of Experiment 9. 152
10 Design of Experiment 10 160
xi List of Figures
Figure Figure title Page
1 A pictorial depiction of the comparator hypothesis. 41
2 A comparison of the ‘orthodox’ model and a revised model of 56
amygdala connections involved in fear conditioning.
3 Similarities and differences in intracellular signaling leading to 78
consolidation of fear conditioning and extinction in the amygdala.
4 Role of IL projections to the amygdala in modulating fear-related 98
neural transmission
5 Neural model of expression and modulation of extinction. 99
6 Mean CS freezing during each stage of Experiment 1. 107
7 Mean CS freezing during each stage of Experiment 2. 115
8 Mean freezing to each CS presentation during test in Experiment 2. 117
9 Mean CS freezing during each stage of Experiment 3. 121
10 Mean CS freezing during each stage of Experiment 4. 127
11 Individual test data points with 95% confidence intervals for 130
population means for groups Control, Over - Vehicle, and Over -
FG7142 from subjects reported in Experiments 2, 3, and 4.
12 Mean freezing to test CS presentations in Experiment 5. 134
13 Mean freezing to test CS presentations in Experiment 6a. 138
14 Mean freezing to test CS presentations in Experiment 6b. 140
15 Mean CS freezing during each stage of Experiment 7. 144
16 Mean freezing in Experiment 8a. 149
17 Mean freezing during test context exposure in Experiment 8b. 149
18 Mean CS freezing during each Stage of Experiment 9. 152
xii 19 Test data from Experiment 8, with group AAA excluded. 156
20 Mean CS freezing during extinction and test in Experiment 10. 160
21 An example of a second-order comparator process. 167
xiii Abbreviations
AB accessory basal (or basomedial) nucleus of the amygdala
AC alternating current
ACFS artificial cerebro-spinal fluid
AMPA α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid
AP5 DL-2-amino-5-phosphopentanoic acid
AP7 DL-2-amino-7-phosphopentanoic acid
BA basal (or basolateral) nucleus of the amygdala
BLA basolateral complex of the amygdala
BOLD blood-oxygen level-dependent cAMP cyclic adenosine monophosphate
CeA central nucleus of the amygdala
CeM medial subdivision of the central nucleus of the amygdala
CI conditioned inhibitor cm centimeter
CNQX 6-cyano-7-nitroquinoxaline-2,3-dione
CPP 3-((+/-)2-carboxypiperazin-4yl)propel-1-phosphate
CR conditioned response
CREB cyclic adenosine monophosphate-response element binding protein
CS conditioned stimulus
CTAP D-Phe-Cys-Tyr-D-Trp-Arg-Thr-Pen-Thr-NH2 dB decibel
DH dorsal hippocampus
ERK extracellular signal-regulated kinase
FG7142 N-methyl-β-carboline-3-carboxymide
xiv FK-506 Tacrolimus fMRI functional magnetic resonance imaging
FPS fear-potentiated startle
GABA γ-aminobutyric acid
GAD glutamate decarboxylase
H3 tritium
Hz Hertz
IL infralimbic cortex i.p. intraperitoneal
IPSC inhibitory post-synaptic current
ISI inter-stimulus interval
ITC intercalated cell mass of the amygdala kg kilogram
LA lateral nucleus of the amygdala
LAd dorsal subdivision of the lateral nucleus of the amygdala
LAv ventral subdivision of the lateral nucleus of the amygdala
LED light-emitting diode
LVGCC L-type voltage-gated calcium channel m meter mA miliampere
MAPK mitogen-activated protein kinase
MeA medial nucleus of the amygdala
MEK mitogen-activated protein kinase kinase mg milligrams
MG medial geniculate body of the thalamus
xv min minute ml milliliters mm millimeter mPFC medial prefrontal cortex mRNA messenger ribonucleic acid
NMDA N-methyl-D-aspartate
PAG periaqueductal gray pCREB phosphorylated cyclic adenosine monophosphate-response element
binding protein
PD098059 2-(2'-amino-3'-methoxyphenyl)-oxanaphthalen-4-one
PFC prefrontal cortex
PI-3 kinase phosphatidylinositol 3-kinase
PKA protein kinase A
PL prelimbic cortex pMAPK phosphorylated mitogen-activated protein kinase
PPF paired-pulse facilitation
Rp-cAMPS Rp-adenosine-3’,5’-cyclic monophosphorothioate s second s.c. subcutaneous
SEM standard error of the mean
SR141716A 5-(4-Chlorophenyl)-1-(2,4-dichloro-phenyl)-4-methyl-N-(piperidin-1-
yl)-1H-pyrazole-3-carboxamide
TTX tetrodotoxin
U0126 1,4-diamino-2,3-dicyano-1,4-bis(2-aminophenylthio)butadiene
US unconditional stimulus
xvi veh vehicle
VH ventral hippocampus vlPAG ventrolateral subdivision of the periaqueductal gray vmPFC ventromedial prefrontal cortex vol volume w weight
WIN 55-212,2 (R)-(+)-[2,3-Dihydro-5-methyl-3-(4-
morpholinylmethyl)pyrrolo[1,2,3-de)-1,4-benzoxazin-6-yl]-1-
napthalenylmethanone
xvii
Introduction
An animal that is unable to predict danger is at an evolutionary disadvantage relative to an animal that can. However, stimuli that signal the spatial or temporal proximity of a dangerous or aversive stimulus or event are not always inherently aversive or dangerous themselves. Thus, animals have evolved the ability to react defensively in the presence of a stimulus (referred to as a conditioned stimulus, or CS) that has previously been paired with an aversive consequence (an ‘unconditioned stimulus’, or US), even when the CS is not inherently aversive. This form of learning is called Pavlovian fear conditioning. While the ability to form CS - US associations is generally adaptive, such learning is also widely believed to underlie the formation of pathological associations that produce inappropriate and disruptive fear reactions seen in some anxiety disorders such as phobias and post-traumatic stress disorder. Indeed, current treatments for these disorders are based to a large extent on what is known of Pavlovian fear conditioning in the laboratory (Ayres, 1998; Garakani, Mathew, & Charney 2006). In addition, the well- established and validated methods for the study of simple forms of learning such as fear conditioning may help to reveal more general rules and processes governing the learning of associations between stimuli, and thus help advance the scientific investigation of learning more generally.
The body of experimental work presented in this thesis is an effort to better understand the neurobiological mechanisms contributing to the expression of overexpectation, a form of learning that leads to reduced conditional fear. This work is, therefore, situated within the context of the broader effort currently underway in the fields of learning theory, neuroscience, and clinical psychology to understand the mechanisms that allow fear extinction to be learned and expressed. In extinction, fear
1
reactions to a CS are diminished after repeated exposures to that stimulus without aversive consequences. The acquisition, consolidation, expression, and extinction of conditioned fear have been extensively studied, answering many questions about its neurobiological substrates.
Contemporary findings and theories regarding extinction suggest that it may involve multiple types of learning, which may engage distinct neurobiological substrates. One type of learning that appears to make a critical contribution to extinction is error correction, whereby changes in CS - US associations are driven by the difference or
‘error’ between the outcome predicted by an organism on the basis of the CS(s) presented and the actual outcome (e.g. Rescorla & Wagner, 1972). In extinction, the actual outcome of CS presentation is the absence of the expected US, creating a negative discrepancy between expectation and outcome, and it is widely believed that detection of this predictive error drives the loss of the conditioned response (CR).
However, given that multiple mechanisms may contribute to extinction learning, it is unclear to what extent prediction error contributes to each particular putative extinction mechanism.
Overexpectation may provide a model that better isolates the contribution of predictive error to decrements in conditioned fear. In a typical overexpectation experiment, two CSs are conditioned separately with the same US such that, by the end of this first stage of conditioning, each CS independently elicits strong conditioned responding. In a second stage of conditioning, the CSs are presented simultaneously, and this compound CS receives further pairings with the same US used in the first stage of conditioning. As will be reviewed later, learning resulting from this compound conditioning can lead to a decrement in the CR to each CS presented alone. If it is to be assumed that CR performance is related to the current associative strength of the present
2
CS(s), then this decrement can only be explained by the operation of prediction error.
According to some models of error-correction, the presentation of the two CSs together may cause summation of the US-expectation elicited by each CS alone. If this degree of
US-expectation is higher than that appropriate to prepare the animal for the US actually delivered, a negative prediction error is produced and, as in extinction, results in decreased CR magnitude upon subsequent presentations of either CS alone.
As the US is present in both training phases, overexpectation may help isolate contributions to decrements in fear that are independent of learning that only occurs in the absence of the US. A popular theory of extinction proposes that extinction involves the formation of a ‘CS – no-US’ memory. The presence of overexpectation, however, suggests the contribution of more general error-correction mechanisms to decrements in fear and, thus, at least in part, to extinction. Overexpectation may be useful, therefore, as a model of specific components of extinction learning that are driven by these mechanisms. Investigating this phenomenon can help contribute to the construction of a more precise understanding of extinction learning and its neurobiological substrates.
3 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Chapter 1
Behavioural characteristics of the learning
and loss of conditioned fear
This chapter will first review early evidence that the pairing of a neutral stimulus with an inherently aversive outcome can lead to profound changes in behavioural responses to subsequent presentations of the CS, indicative of fear conditioning. Early demonstrations of CRs now commonly used to measure conditioned fear will be reviewed. Following this, the importance of predictive error to the acquisition of conditioned fear will be discussed. This discussion will focus on experiments examining the blocking effect and the implications of these experiments regarding the nature of the prediction errors that control learning. Similarly, evidence that extinction depends on predictive error, as suggested by the ‘protection from extinction’ effect will be discussed. The purpose of discussing the role of prediction error in learning and loss of fear is to highlight the fact that the outcome (or lack thereof) of a conditioning trial is insufficient to produce changes in conditioned associations. Prior associations, and the outcome-expectations they produce, determine whether reinforcement or non- reinforcement of a stimulus results in associative change. Understanding this is important for understanding overexpectation, and its relevance to the study of extinction.
Empirical data will then be reviewed relating to phenomena that suggest that, despite the loss of conditioned responding, CS - US associations survive extinction. These data are important as they have influenced the development of several theories regarding the content of extinction learning, which will be discussed. Indeed, some recent data 4 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
suggests that behavioural extinction may be the common result of several very different mechanisms, which may be differentially engaged in different extinction training circumstances. Again, this points to the relevance of overexpectation for developing a more precise understanding of the different components of extinction learning. Thus, having discussed several current theories of extinction, attention will be turned to empirical data regarding overexpectation, and its theoretical implications. By necessity, the review of the literature will be selective, focusing mainly on investigations of fear conditioning and only drawing on evidence from other conditioning preparations where necessary.
1. Acquisition of Pavlovian Fear Conditioning
Conditioning of a fear reaction was first reported by Watson and Watson (1921), who describe a case study involving an 11 month old boy, “Albert B”. The experimenters observed that Albert responded to the sound of a steel bar being struck with a hammer behind his head with changes in respiration and facial expression, flailing of the arms, and, on some trials, hiding his face, crying, and attempting to flee by crawling away from the source of the noise. Thus this loud noise was an effective fear US. It was also observed that, on its first presentation, a white rat elicited reaching from Albert, suggesting that his unconditional response to the rat was curiosity rather than fear. Conditioning trials were presented in which presentation of the rat was followed by the loud noise at the point at which Albert’s hand made contact with the rat.
After two such conditioning trials, Albert became reticent to touch the rat. After an additional three trials, presentation of the rat elicited whimpering and Albert turned away from the rat. Finally, after an additional 2 trials (7 in total), presentation of the rat caused Albert to instantly begin crying and crawling away from the rat rapidly.
5 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Of course, the study described above, being an uncontrolled case study, does not permit a firm conclusion regarding whether the change in Albert’s behavioural response to the rat was due specifically to the rat - noise pairings. The behavioural change may have reflected a more general sensitisation of fear responses to the rat, or simply reflected peculiar behaviour patterns specific to Albert. Experimental studies employing animals exposed to control conditioning procedures, in which the CS and/or US are either not presented or presented in an unpaired manner during the conditioning session, have helped examine the degree to which pairing of a CS and US is necessary for observation of a conditioned response. In doing so, a number of quantifiable, well- defined measures of conditioned fear have been developed.
One of the most popular ways to measure conditioned fear in the laboratory has been to observe the degree to which presentation of a CS leads to suppression, relative to a pre-CS baseline, of reward-seeking behaviour. For example, rats suppress food- reinforced bar-pressing to a tone after it has been paired with shock (Estes & Skinner,
1941). Rats exposed to conditioning in which a tone co-terminates with shock show greater suppression of drinking in the presence of the tone than rats exposed to either backward conditioning, in which shock ended with CS onset, trace conditioning, in which shock presentation was delayed until 10-60 s after tone offset, or exposed only to the CS, only to the US, or to neither stimulus (Leaf & Muller, 1965; Leaf & Leaf,
1966). One of the direct causes of fear CRs such as suppression of bar-pressing and drinking in rats is the dramatic decline in all activity often seen in rats and mice when encountering a feared CS. This is often expressed as freezing, a species-specific defence response in which rodents adopt a tense, crouching posture and cease all movements aside from those required for breathing. Bouton and Bolles (1980) confirmed the validity of using this measure as a measure of conditioned fear by showing that it is
6 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
correlated with conditioned suppression of operant bar-pressing, drinking, and exploratory behaviour in rats conditioned with forward pairings of a 60 s tone with 0.5 s, 1 mA shock. Freezing was also consistently higher in forward-conditioned rats than in rats exposed to backward conditioning.
Fear can also be measured by its ability to potentiate other fear- and anxiety-related responses. For example, Brown, Kalish, and Farber (1951) reported that after pairings of a 5 s light - buzzer compound stimulus with a 2 s, 0.21 – 0.24 mA shock, rats showed larger startle jumps to the sound of a toy pistol being fired during test trials of the CS than they had shown prior to conditioning. These startle jumps were also larger than in rats exposed to an unpaired ‘pseudo-conditioning’ schedule in which US presentations either preceded or followed the CS by 3 s. Pseudo-conditioning elicited no increase in startle amplitude in the presence of the CS relative to pre-conditioning levels.
In laboratory studies using humans as subjects, researchers have often taken advantage of changes in psychophysiological measures, rather than overt behaviour, to index conditioned fear. For example, Mordkoff, Edelberg, and Ustick (1967) exposed male students to 6 trial blocks, each involving 5 pairings of an 8 s tone (the CS+) with
0.5 s shock. Changes in skin conductance in the palmar and dorsal surfaces of the hand were measured during these trials and classified as orienting responses to the tone (1 –
2.9 s after CS onset), anticipatory responses (3 - 8.5 s after CS onset), and unconditioned responses to shock (8.6 - 10.5 s after CS onset). The orienting response declined over the course of the conditioning trials, but this did not reflect conditioning because these responses did not differ from those shown to a tone of different pitch (the
CS-) that was presented 3 times per trial block. The anticipatory response to the CS+, however, increased significantly over the trial blocks, and was significantly higher than that shown to the CS-. When comparing responses to non-reinforced CS+ presentations,
7 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
two of which were presented during each trial block, with responses to the CS-, differences were also observed during the 8.6 – 10.5 s post-CS onset epoch, perhaps reflecting differential expectancy violation.
1.1. The role of error-correction in the acquisition of conditioned fear.
The early studies of fear conditioning described above confirm that contiguity of the
CS and US can lead to learning of conditioned fear responses. However, such contiguity is insufficient for associative learning. Acquisition of CS - US associations can be modulated, and even prevented, depending on the degree to which the US is or is not expected. Several models of associative learning describe the effects of US prediction on acquisition of CS - US associations through error-correction rules. When an unfamiliar stimulus which has no inherent aversive or appetitive motivational properties is first presented to an animal, it is associatively neutral, as the animal has no prior experience of it being correlated with any other form of stimulation. It therefore elicits no prediction of a motivationally significant outcome. This lack of prediction is an error if the stimulus is then paired with a US. Associative learning corrects this error so that on subsequent presentations of the CS, the animal may make appropriate responses to prepare for the predicted US.
Of course, associative learning is often unlikely to proceed in a purely binary fashion, where the animal simply anticipates either the presence (1) or absence (0) of a
US and adjusts CS - US association to the opposite value if its prediction proves incorrect after a single trial. In nature, a US may be present at different intensities in different instances. Its pairing with certain stimuli or combinations of stimuli may be more or less consistent. It may therefore be adaptive for an animal to be able to ‘fine tune’ the magnitude or quality of its response to a CS after initial conditioning trials.
8 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Furthermore, due to various properties of the CS and US (e.g. salience, magnitude), the
CS - US relationship (e.g. degree of temporal overlap, consistency of reinforcement), the animal (e.g. arousal, sensory acuity) and other stimuli present (e.g. relative salience, conditioned or unconditioned motivational properties), the degree to which learning from any single CS - US pairing corrects a prediction error may vary widely.
The Rescorla-Wagner model of Pavlovian association formation (Rescorla &
Wagner, 1972) accounts for the influence of several factors over the degree of change of associative strength resulting from a CS - US pairing with the equation;
ΔV = αβ(λ − ΣV) (1)
in which V represents the strength of a CS - US association, α a learning rate parameter determined by the properties of the CS, β a learning rate parameter determined by the properties of the US, and λ the maximum associative strength that can be conditioned by the US. Thus, Equation 1 states that changes in associative strength resulting from a
CS - US pairing are determined by the difference between the maximum associative strength that can potentially be conditioned by a US and the pre-existing associative strength of the CS and other present stimuli. The proportion of this error corrected on each individual pairing is determined by specific properties of the CS and US.
The ‘blocking’ effect is commonly cited as evidence for the role of prediction error in fear conditioning. Blocking was demonstrated by Kamin (1968) in rats conditioned to associate a 3 min CS (‘A’: either a light or an 80 dB noise) with a 0.5 s, 1 mA shock over 16 CS - US pairings. An additional 8 presentations of the CS, in compound with a new CS (‘B’: noise or light, depending on which CS had been conditioned first) that had not received any previous conditioning, were then paired with the same shock.
9 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Suppression of food-reinforced bar-pressing was then measured in the presence of CSB alone. Rats in this ‘blocking’ group were compared with rats that had either received only the compound - shock pairings, or 24 pairings of only CSA with shock, or the compound and CSA-shock phases described for the experimental group in reverse order.
Regardless of whether the light or noise served as the test stimulus, CSB elicited less suppression in the blocking group than in the reverse-order and compound-only groups.
Indeed, suppression to CSB in the blocking group was equivalent to that seen in rats that had only received the first stage of conditioning of CSA, and which had thus never experienced pairing of CSB with shock. This suggests that, in the blocking group, the pre-existing associative strength of CSA blocked CSB from entering into association with the shock during the compound - shock pairings. Consistent with the idea that the intensity of the US sets an upper limit on associative strength that can be conditioned, increasing the intensity of the shock from 1 mA during conditioning of CSA to 4 mA during compound conditioning abolished this blocking effect. According to the
Equation 1, this increase in US intensity would increase the value of λ. Thus, even if
CSA had acquired associative strength equal to the λ value of the 1 mA shock during the first stage of conditioning, λ − ΣV would again have a positive value during compound conditioning, allowing acquisition of associative strength to CSB.
Some theorists have proposed that blocking results not from the absence of predictive error generated by a pooled error term, but by learned inattention to CSB, based on its predictive redundancy. Mackintosh (1975a) proposes that in a conditioning trial, each CS present can have its own error term. This is expressed in the equation:
ΔVA = αAβ(λ − VA) (2)
10 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Therefore, increments in associative strength can occur to each CS independently of those occurring to other CSs present. In Equation 2, αA represents attention to stimulus
A. Unlike in Equation 1, where αA is a constant, in Mackintosh’s (1975a) model, αA can change from trial to trial depending on stimulus A’s predictive validity relative to other stimuli present. If CSA has been established as a reliable predictor of a US, its α value will be maintained at a relatively high level. When a new stimulus, CSB, is introduced in compound with CSA, it will have normal associability on the first trial. However, given the presence of a better predictor, CSA, αB would decline over subsequent pairings. Thus, even if the error term of CSB (λ − VB) remains high, learning resulting from it will be low due to its multiplication by a low, and ever-decreasing, αB term.
An important point where US-processing vs. CS-processing accounts of blocking differ is in their prediction of the presence vs. absence of one-trial blocking. If only 1 compound - shock pairing is presented, an animal has no chance to learn to down- regulate attention to CSB during subsequent reinforced compound trials occurring prior to test. Therefore, according to Mackintosh (1975a), 1 compound - shock pairing should not produce a blocking effect. The increment in associative strength between CSB and the US on the first trial should be just as high as in animals that did not receive conditioning with CSA before compound conditioning. Blocking can only develop across further compound - shock presentations as the ability of CSB to acquire further associative strength declines. According to Rescorla and Wagner (1972), however, the presence of a feared CSA should prevent an associatively neutral CSB from entering into association with the shock on any trial, including the first compound - shock pairing. Using a conditioning procedure that produced blocking when 8 compound - shock pairings were presented, Mackintosh (1975b) found that, when given only 1 compound trial, the degree of conditioned suppression of licking in water-deprived rats 11 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
to CSB on test was same regardless of whether or not CSA was conditioned prior to the compound trial. Suppression to CSB in these groups of rats was greater than in a generalisation control group that received only stage 1 conditioning of CSA, showing that some CSB-shock association was acquired. These results are consistent with the suggestion that the blocking effect is caused by changes in CS processing that require more than one compound - shock pairing to develop.
However, this conclusion is disputed by the findings of Rescorla (1970), in an experiment which, according to Mackintosh (1975a), would have equated the α values of CSA and CSB during compound conditioning. In this experiment, rats were subjected to 3 pairings each of a flashing light and a tone with shock, which caused sub- asymptotic conditioning to both stimuli. Further conditioning of the compound of the tone and light caused no increase in conditioned suppression to either stimulus when subsequently tested alone, compared to rats that only received the first stage of conditioning. Groups of rats that received further conditioning of one of the stimuli alone showed greater suppression to that stimulus on test, showing that the blocking observed could not be attributed to ceiling effects. As the two stimuli had identical conditioning histories prior to compound conditioning, their α values would have been identical during compound conditioning. Therefore, the failure to show further conditioning cannot be attributed to learned inattention to a less valid predictor. If the stimuli could each gain associative strength independently of the other’s presence and associative strength, compound conditioning should have continued to enhance their ability to elicit suppression of food-reinforced bar-pressing.
This result is, however, consistent with the summation of associative strengths of the two stimuli preventing further acquisition. Furthermore, in contrast to Mackintosh’s
(1975b) negative finding, Cole and McNally (2007) demonstrated one-trial blocking in
12 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
a within-subjects design using freezing in rats as a measure of conditioned fear. In this study, a CSB, conditioned in compound with a pre-conditioned CSA could be blocked relative to a CSD, conditioned in compound with another novel stimulus CSC, after only 1 compound - shock pairing. Thus while attentional, CS processing mechanisms may contribute to blocking in at least some conditioning preparations, it appears that US processing mechanisms are often as important, if not more important, contributors.
A theoretically important point suggested by the blocking phenomenon is that mere temporal contiguity of a CS and US is not a sufficient determinant of the formation of an internal CS - US association. Formation of new associative strength requires that the
US not be already predicted on the basis of stimuli present. This can be referred to as a
‘positive’ prediction error. Error-correction theories also postulate that prediction error can be ‘negative’ when the magnitude of the US is lower than what is expected on the basis of present stimuli. These theories predict that the consequences of negative prediction error are the symmetrical opposite of positive predictive error: decrements in
CRs. However, empirical evidence regarding extinction and overexpectation suggests that loss of fear involves more than a simple reversal of what is learned during extinction. This evidence, and its theoretical implications, will be discussed in the following sections of this chapter. In the case of extinction, this review will, by necessity, focus on studies of fear conditioning, referring to studies of other preparations only where necessary. For a comprehensive review of extinction of appetitive conditioning, see Delamater (2004).
2. Extinction of Pavlovian fear conditioning
Extinction refers to the loss of conditioned responding to a CS after it is presented
(usually multiple times) in the absence of the reinforcer that it was previously associated
13 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
with. Extinction, like acquisition, depends on prediction error, which produces decrements in CS - US association that translate into decrements in fear CRs. However, the expression of extinction memory appears to be generally less stable than memory for conditioning. Extinction is often observed to be attenuated when test occurs outside the context of extinction training, after re-exposure to the US used in conditioning, or after an extended period of time. These findings have led many to propose that extinction involves the formation of a new memory with many distinct properties from that formed during acquisition, possibly relying on very distinct mechanisms.
2.1. The contribution of error correction to extinction.
Equation 1 models extinction by setting λ = 0, to signify the absence of a US. If a
CS has a positive association with a US (V > 0), then the absence of reinforcement causes the parenthetical term of Equation 1 to take a negative value: (λ - ΣV) < 0. This negative prediction error generates a downward correction in associative strength. This loss of associative strength renders the CS less able to elicit conditioned responding on later presentations. Rescorla and Wagner (1972) allowed the learning rate parameter of non-reinforcement, β, to take a different value to that associated with reinforcement by the US. This accommodates the common observation that extinction of a fear CR is often learned much more slowly than acquisition of the CR. This modification, however, does not alter several important predictions about extinction made by this model. These include the predictions that an extinguished CS should be incapable of blocking acquisition of conditioned responding to a second stimulus, and that extinction of a CS should be attenuated by the presence of a conditioned inhibitor.
To test the ability of an extinguished CS to block acquisition of fear to a second CS in rats, Kamin (1968) paired a noise with shock four times and then presented the noise
14 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
without shock an additional 12 times. A noise-light compound was then reinforced with shock an additional eight times. Following this, the ability of presentations of the light to suppress food-reinforced bar-pressing was measured. Suppression in these rats was significantly stronger than in rats that had not received the non-reinforced noise presentations between conditioning of the noise and conditioning of the compound, and was not significantly different to that shown by rats only exposed to compound conditioning. Thus, extinction of the noise prior to compound conditioning abolished its ability to block conditioning to the light.
A further prediction of Equation 1 is that extinction can be blocked if extinction training is conducted in the presence of stimuli that have negative associative strength.
Negative associative strength can be achieved by conditioned inhibition, where one stimulus, CSA, is paired with a US when presented alone, and is non-reinforced whenever presented in compound with a second stimulus, CSB. CSA gains positive associative strength with the US, meaning that its non-reinforcement (λ = 0) in the presence of CSB results in a negative prediction error, because, when CSB is neutral,
VB = 0 and, therefore, ΣV = VA. This generates a negative decrement in associative strength which is distributed between CSA and CSB according to their relative α values.
CSB, having never gained positive associative strength, will thereby gain negative associative strength. Such a stimulus is referred to as a conditioned inhibitor (CI). It is worth noting that, while Equation 1 provides an elegant account of the acquisition of conditioned inhibition, it also predicts that simple non-reinforced presentations of a CI should cause loss of conditioned inhibition. However, numerous investigations have failed to show such an effect (e.g. Zimmer-Hart & Rescorla, 1974).
An example of protection of a CS from extinction by co-presentation of a CI comes from Soltysik, Wolfe, Nicholas, Wilson, & Garcia-Sanchez (1983), who conditioned
15 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
cats to fear 2 CSs through a trace conditioning procedure in which offset of the CS was followed by a 0.3 s, 2 - 2.5 mA shock after an interval of 3 s. On trials on which another
3 s stimulus immediately followed a CS, the shock was not presented, thus conditioning the second stimulus as a putative CI. During extinction training, one of the CSs was presented alone, while the other was always followed by the CI. On test, the second CS elicited a greater conditioned suppression of respiratory amplitude than the one extinguished alone. This is consistent with the Rescorla-Wagner rule, which predicts that if the negative associative strength of a CI and the positive associative strength of a
CS summate to produce no net US prediction, then no prediction error occurs during an extinction trial. Thus, no associative change should occur for either stimulus.
However, because Soltysik et al. (1983) did not include a CS extinguished in the presence of a control stimulus that was not a CI, it is unclear whether the effect they detected results specifically from the inhibitory properties of the co-present stimulus, or more generally from the presence of any extra stimulus during conditioning promoting a generalisation decrement. Indeed Lovibond, Davis, and O’Flaherty (2000) found, in human fear conditioning, that even a conditioned excitor could also effectively protect a
CS from extinction. This is contrary to the prediction made by Equation 1 that the presence of a conditioned excitor should enhance extinction. Indeed, using several appetitive conditioning designs in rats, Rescorla (2000) found opposite results to those of Lovibond et al. (2000). Extinction training of either a Pavlovian or instrumental response in the presence of an additional appetitive excitor enhanced, rather than impaired, the reduction in response as measured during a subsequent test. These findings question the generality of Lovibond et al.’s (2000) finding of protection from extinction by a concurrent excitor. Furthermore, Rescorla (2003) has found protection from extinction to be greater when the protector stimulus is a CI than when it is not.
16 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
When the control stimulus’ only associative history was reinforcement in compound with the same CS used to condition inhibition to the CI (in other words an associatively blocked stimulus), it was less effective than the CI at protecting conditioned key- pecking in pigeons or magazine entries in rats from extinction. A stimulus that had previously been presented but never paired with any other CS or US was also less effective than a CI at protecting a CS from extinction of auto-shaping in pigeons.
Demonstrations of protection from extinction are consistent with the implication of
Equation 1, that predictive error plays just as important a role in the loss of conditioned responding as it does in its acquisition. Just as the predictability of a US reduces its ability to condition new associative strength, the predictability of its non-occurrence reduces extinction of established associations. The failure of an extinguished CS to block conditioning to a second stimulus is, furthermore, consistent with the absence of
US prediction after extinction. This could be seen as indicating that extinction erases the
CS - US association. However, numerous demonstrations of the return of conditioned responding after extinction have been cited as evidence that the CS - US association in some way survives extinction. These include several ‘signature phenomena’ of extinction; spontaneous recovery, where extinguished CRs may reappear after passage of time; context-mediated renewal, where extinguished CRs are stronger in contexts other than that in which they were extinguished; and reinstatement, where extinguished
CRs reappear after unsignalled presentations of the US.
2.2. Reinstatement
Reinstatement of conditioned fear was first described by Rescorla and Heth (1975), who conditioned rats to fear a 2 min tone with a 0.5 s, 3 mA shock. After 20 non- reinforced post-conditioning presentations of the tone over 5 days, rats were re-exposed
17 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
to a single presentation of the shock. When their reactions to the tone were tested the following day, these rats showed greater suppression of food-reinforced bar-pressing than rats that did not receive the reinstating shock or rats that received the shock after a conditioning schedule involving unpaired shock and tone presentations. Neither of these control groups differed in suppression on test from a group conditioned with the unpaired procedure and not presented with a reinstating shock, giving no suggestion of a role for post-shock sensitisation in the reinstatement effect.
Rescorla and Heth (1975) interpreted this effect as resulting from the restoration of the strength of an internal US representation that had been degraded by extinction learning. According to this account, extinction leaves the CS - US association at least partly intact, but the US representation that is excited by the CS is somehow degraded in its ability to elicit a central state of fear and/or its behavioural correlates. Re-exposure to the shock restores the behavioural potency of this internal US representation.
Rescorla and Heth (1975) propose that the aspects of the US representation reinstated by the post-extinction shock include its general affective valence, as loud noises could at least partly reinstate fear to a tone conditioned with shock, and vice versa.
However, further investigation of the reinstatement phenomenon has tended to support an interpretation of the effect that appeals to contextual learning caused by the post-extinction shock. Exposure to unsignalled shock in the same context as that used for extinction and test promotes superior reinstatement of conditioned suppression to a tone CS than shock exposure in a separate context (Bouton 1984). This context- dependency of reinstatement is also seen in skin conductance responses in humans conditioned to associate a visual CS with an uncomfortably loud (100 dB) noise burst.
CRs tested in the extinction context were greater if post-conditioning re-exposure to the
18 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
loud noise was conducted in the same context than if conducted in a different context
(LaBar & Phelps, 2005).
Westbrook, Iordanova, McNally, Richardson, and Harris (2002) found that, if extinction and test contexts differed, unsignalled shock was only effective at reinstating freezing in rats if it occurred in one of these two contexts. The effect of unsignalled shock in an extinction context was specific to stimuli extinguished in that context, as freezing to stimuli paired with the same US but extinguished in another context did not show reinstatement when tested in a novel context after unsignalled shock. Neither was there substantial evidence for reinstatement on test if the extinction/reinstatement context received further extinction training after the reinstating shock. Westbrook et al.
(2002) concluded that reinstatement could actually result from two different forms of contextual learning. One is mediated conditioning, where an internal representation of a
CS, elicited by exposure to a context that has been paired with that CS, can itself associate with a shock presented in that context. Another is a context-US association that helps retrieve the latent CS - US association of any stimulus tested in that context.
2.3 Renewal
The return of conditioned responding when a CS is presented outside a context in which it has been extinguished is a very robust effect observed across virtually every conditioning preparation where it has been investigated (Bouton, 2004). In humans treated for spider fear by exposure therapy, self-reports of fear during a spider exposure session one week after treatment were greater when tested in an unfamiliar context than when tested in the same context as treatment was conducted (Mineka, Mystkowski,
Hladek, & Rodriguez, 1999; Mystkowski, Craske, & Echiverri, 2002). In rats, conditioned with tone-shock pairings and exposed to the tone without reinforcement in
19 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
a second ‘B’ context, suppression was higher when the tone was tested in the conditioning context (‘ABA’ renewal) or a third, associatively neutral context (ABC renewal), than when tested in the extinction context (an ‘ABB’ control condition;
Bouton & Bolles, 1979).
The Rescorla-Wagner model could account for these effects by assuming that the extinction context, as a background stimulus to the feared CS, becomes a fear CI.
According to this account, the negative associative strength of the context protects some of the CS’s positive associative strength from extinction. When the CS is presented in a context that does not have inhibitory properties, this remaining associative strength is manifest. However, by arranging for extinction and test contexts to have equal histories of CS reinforcement and non-reinforcement, it has been shown that renewal cannot simply be explained by inhibitory conditioning of the extinction context (Bouton &
Ricker, 1994; Harris, Jones, Bailey, & Westbrook, 2000).
Bouton and Ricker (1994) used an AAB vs AAA design in which a noise and light were each paired with shock in different contexts. Each CS was then extinguished in its conditioning context. When tested, the light elicited greater conditioned suppression of food-reinforced bar-pressing in the context used for noise training than in its own training context. Similar results were found by Harris et al. (2000), using an ABC vs
ABB design in which conditioning of a clicker and a white noise with shock was conducted in context A, and then extinction of them conducted in two separate contexts, each different from the conditioning context. When tested in its own extinction context, a stimulus elicited less freezing than if tested in the context in which the other stimulus had been extinguished.
These findings appear to argue against renewal being due to the conditioned inhibitory properties of an extinction context. However, they still could be
20 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
accommodated within a model of renewal that relies on the extinction context functioning as a CI if it is assumed that transfer of the context’s inhibitory power to an excitor other than that used to condition the context as an inhibitor is incomplete.
Indeed, using discreet stimuli, Zimmer-Hart and Rescorla (1974) found evidence that such transfer of conditioned fear inhibition is incomplete. They exposed rats to pairings of both a tone and a clicker with 0.5 s, 0.5 mA shock, designed to condition both stimuli as fear CSs, along with non-reinforced compound presentations of the clicker and a flashing light, designed to condition the visual stimulus as a fear CI. When tested after this procedure, the light-clicker compound elicited less suppression of food-seeking behaviour than the light-tone compound, suggesting that the light was a less potent CI in the presence of the tone, with which it had not been previously presented in compound, than in the presence of the clocker that was used to condition it as an inhibitor.
The lack of counter-balancing of stimuli in Zimmer-Hart and Rescorla’s (1974) investigation casts doubt on whether their results reflect an actual lack of transfer of inhibition or whether they simply reflect inherent differences in rats’ responding to a tone vs. a clicker. However, in experiments investigating autoshaping in pigeons that did counterbalance stimuli, Rescorla (1982) confirmed that transfer of inhibition to an excitor other than that used to condition inhibition to the CI was incomplete.
Furthermore, it appeared that this impairment of transfer was due to within-compound associations formed between the CS and CI during conditioning of inhibition, as non- reinforced presentations of the CS used to condition inhibition to the CI between training and test improved transfer of inhibition. Thus, Bouton and Ricker’s (1994) and
Harris et al.’s (2000) findings on renewal may be due to conditioned inhibitory properties of extinction contexts. Though extinction and test contexts in both studied would have been matched for any inhibitory properties acquired during extinction
21 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
sessions, context-CS associations formed during extinction training may have led to contexts functioning as less potent inhibitors of a CS extinguished elsewhere than of a
CS extinguished in that context.
In addition to shifts in physical context between extinction and test, shifts in internal drug state have been shown to induce renewal. Bouton, Kenney, and Rosengard (1990) paired a context with shock, and then extinguished the context 30 min after an intraperitoneal (i.p.) injection of chlordiazepoxide, diazepam, or vehicle. Both benzodiazepines caused dose-dependent amnesia for extinction training when rats were tested in a drug-free state. If rats extinguished under the influence of chlordiazepoxide were tested in the same drug state as they received extinction training, context exposure caused less suppression of drinking than in non-extinguished controls and in rats tested drug-free. Thus, expression of extinction learning appears to sometimes be highly specific to both the external contextual states and internal pharmacological states in which it was learned.
2.4 Spontaneous Recovery
The spontaneous reappearance of an extinguished CR after passage of sufficient time is a phenomenon widely reported throughout the history of the study of both Pavlovian and instrumental conditioning, and has been observed in a wide variety of conditioning preparations (Rescorla, 2004). One example of a study of spontaneous recovery in
Pavlovian fear conditioning comes from Quirk (2002), who exposed rats to seven tone- shock pairings and then presented, after a 1 hour delay, a further 20 tone presentations without reinforcement. Rats’ freezing to the tone was then tested 30 min or 1, 2, 4, 6,
10, or 14 days after this extinction session. The longer the delay between extinction and test, the more rats froze, indicating less recall of extinction. At 10 and 14 days, rats
22 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
freezing was statistically indistinguishable from their freezing during the first two extinction trials. Though this suggests complete loss of the extinction memory, rapid re- extinction during the test trials in rats tested 14 days following extinction suggested the preservation of an extinction memory despite it not being initially expressed.
2.5 Theories of extinction
Reinstatement, renewal, and spontaneous recovery are widely considered evidence against the implication of Equation 1, that extinction simply involves the loss of associative strength gained during acquisition. Instead, several prominent theories of extinction propose that it involves new associative learning, the expression of which competes with or inhibits expression of the CS - US association. According to Bouton
(1993) and Pearce and Hall (1980), this inhibitory learning involves the encoding of a representation of “no US”. Thus, an association is learned during extinction training between the CS and this “no US” representation. Bouton’s (1993) account of the role of context in influencing retrieval of a CS - US or a CS – “no US” memory after extinction has become particularly popular among those investigating the behavioural and biological mechanisms of extinction in recent years. According to this model, the memory formed as the result of a conditioning session includes information about the
CS, US, and context. Once stored, the CS - US association remains available indefinitely for future retrieval. Retrieval can, however, be modulated by the similarity or difference between conditions present at the time of retrieval and conditions present at the time of conditioning.
The CS – no-US memory formed during extinction also includes contextual information. Once stored, this extinction memory is also permanently available for retrieval. Thus, after extinction, a CS is ambiguous, having two meanings. Performance
23 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
to a later presentation of the CS is determined by the extent to which the context signals the retrieval of each meaning. This is most simply illustrated in the case of ABA vs
ABB renewal. Context A is associated with the conditioning and therefore signals the
CS - US association. Context B is associated with extinction, and therefore signals the
CS – no-US memory. If test is conducted in a third context, C, the likelihood of renewal will depend on the relative similarity of context C to contexts A and B. This allows
ABC renewal to be a less robust effect than ABA renewal. However, Bouton’s (1993) theory predicts that renewal should generally be detectable whenever there is some shift in context discernable to the subject between extinction and test. This is because Bouton
(1993) allows that biologically significant excitatory memories, such as the pairing of a
CS with a painful US, generalise more widely across contexts than inhibitory memories.
Thus, given a sufficiently strong CS - US memory, involving a sufficiently motivationally significant US, both ABC and AAB renewal should, according to this theory, be robust.
Bouton (1993) suggests that spontaneous recovery may simply be a form of renewal.
Changes in internal hormonal and neurochemical states over time, and changes in external contextual variables such as recent events, gradually change the context. Thus, even if conditioning, extinction, and test is conducted in the same location, with the same objects present, any sufficiently long extinction-test interval should cause renewal.
Reinstatement which is observed after US presentation in the test context is also explained as resulting from mechanisms essentially identical to those involved in renewal. Bouton (1993) suggests that a reinstating US causes contextual learning which renders the animal’s internal representation of the test context more similar, by virtue of its associative properties, to the internal representation of the conditioning context than that of the extinction context. It is less clear, however, how this model accounts for
24 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
reinstatement effects seen when US delivery occurs in an extinction context and test is conducted in a separate context (Westbrook et al., 2002).
However, to the extent that contexts may gain occasion-setting properties based on the history of a CS’s reinforcement, this is probably not the exclusive determinant of contextual control over extinction recall. Denniston, Chang, and Miller (2003) showed that when rats received 800 extinction trials of a noise CS previously paired with shock, both ABC vs. ABB and ABA vs. AAA renewal of lick suppression were attenuated relative to rats presented with 160 noise presentations during extinction. If occasion- setting was the sole determinant of a context’s ability to modulate expression of extinction, greater CS extinction in context B should enhance the occasion-setting properties of that context without interfering with the ability of other contexts to trigger recall of the fear association. Denniston et al. (2003) propose instead that the inhibitory association learned during extinction simply has a narrower generalisation gradient than the excitatory learning, but that this generalisation gradient can be gradually increased with prolonged strengthening of the inhibitory memory.
A second theory regarding the content of new learning resulting from extinction equates the processes mediating extinction to those mediating conditioned inhibition.
While similar to Bouton’s (1993) theory in its implication that an extinguished CS has two separate associations, it takes the view that the content of the second memory formed is simply ‘inhibitory’ rather than ‘no US’, and is thus more generally applicable to losses in conditioned responding that may occur even in the presence of US reinforcement. According to this theory, an extinguished CS is both a conditioned excitor and conditioned inhibitor of the US representation, and when it is presented, its positive and negative associative strengths are both activated. The summation of these
25 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
associations produces a negligible net associative strength, and conditioned responding is not observed.
One account for how this inhibitory association is learned is provided by Wagner
(1981). This account assumes that relationships between stimuli can only be learnt when their internal representations are in an ‘active state’, analogous to working memory.
This active (‘A’) state can be divided into two levels; a focal A1 state and a peripheral
A2 state. Elements of a stimulus’ internal representation can only be activated to A1 if it is actually presented, while associatively activated internal representation elements only enter the A2 state, from which they are unable to enter A1. Thus, when a CS is presented, its representation is in A1, and associatively-activated elements of the US’s representation are in A2. To the extent that elements of a stimulus representation are in
A1 simultaneously with the activation of another stimulus’ representation in A2, an inhibitory association is learnt between the first and second stimulus. However, it is unclear how this model of extinction can account for renewal and spontaneous recovery without additional assumptions regarding the properties of the inhibitory memory.
The theories of extinction discussed so far in this section all explain extinction in terms of modified associations between internal representations of the CS and US.
These internal representations may be unchanged, but the connections between them are weakened, or new connections are formed during extinction that compete with the older connections for behavioural expression. However, some have proposed that changes in the internal representations of either the CS or US themselves are at least partly involved in extinction. The proposal by Rescorla and Heth (1975), that extinction involves the degradation of a US representation, has already been discussed. This model suggests commonalities between extinction and devaluation, a decrement in a CR seen if, after conditioning, a loud noise US is rendered less aversive through habituation in
26 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
rats (Rescorla, 1973) or presentation at lower magnitudes to humans (Hosoba, Iwanaga,
& Seiwa, 2001). In such situations, the CS - US association may be completely intact, but the US representation itself has lost the ability to elicit a emotional and behavioural responses.
Reduced processing of, or attention to, the CS, or reduced ability of CS presentation to excite its own internal representation, may be another mechanism by which repeated presentations of a CS in the absence of reinforcement lead to a decline in conditioned responding. According to this view, extinction and habituation should share common mechanisms and features. McSweeney and Swindell (2002) review extensive evidence that extinction and habituation share many common features. For example, spontaneous recovery of unconditioned responding to a stimulus after habituation has been reported to increase with increased time between habituation and test, just as spontaneous recovery of conditioned responding increases with time since extinction. Repeated habituation or extinction training both reduce spontaneous recovery and may increase the rate of re-habituation or re-extinction after spontaneous recovery or, in the case of extinction, re-conditioning.
Renewal, and renewal-like effects, such as disinhibition, where presentation of a novel or extraneous stimulus before test of an extinguished CS leads to a renewed CR, have also been observed after habituation. Changes in context have been found to restore habituated responding in some preparations. Dishabituation has also been observed after introduction of a novel or extraneous stimulus, which could be seen as analogous to a contextual alteration. Finally, a meta-analysis of studies that provided data regarding within-session changes in responding during extinction or habituation showed that a negative exponential function fits both the extinction and habituation data very well across a range of vertebrate species (McSweeney & Swindell, 2002).
27 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
In recent years, discoveries of different neurobiological substrates for extinction, consistent with different theories of extinction, have encouraged the idea that several, or even all of these theories of extinction may be valid. Contributions to extinction by different mechanisms may depend on numerous variables. Variables of the subject such as species, integrity of certain brain regions, and internal state may create a biological
‘background’ that favours certain mechanisms over others. Variables of the conditioning procedure such as the modality, magnitude, and other qualities of the CS and /or the US, the temporal gap between conditioning and extinction training, and variables of the extinction session such as trial duration, trial spacing, and context may also influence which mechanisms are engaged.
Consistent with the influence of such variables over the relative contributions of different mechanisms to extinction, a recent study has shown that vulnerability of extinction to reinstatement, renewal, and spontaneous recovery is dependent on the conditioning-extinction interval. Myers, Ressler, and Davis (2006) conditioned rats to fear a 3.7 s light by pairing it with a 0.5 s, 0.4mA shock over 15 trials. Following this,
90 extinction trials of the light were presented, beginning either 10 min or 1, 24, or 72 hours after the completion of conditioning. The vulnerability of the extinction learned from these trials to reinstatement was tested by exposing rats to 5 unsignalled shocks 8 days after the initial conditioning session and testing FPS to the light 24 hours later.
This was compared to FPS that had been tested 24 hours after the initial extinction session. Only rats given extinction exposures 72 hours after conditioning exhibited a significant increase in FPS between the first post-extinction test and the post- reinstatement test. In a second experiment, an extinction session 72 hours after conditioning in a second context allowed for renewal of FPS when tested in the conditioning context 24 hours after extinction compared to a group receiving all
28 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
sessions in the same context (ABA vs. AAA renewal). However, the shorter conditioning-extinction intervals yielded no evidence of renewal. In a third experiment, rats exposed to the extinction session either 10 min, 1 hour, or 72 hours after conditioning were tested for FPS either 1 or 21 days after extinction. Significantly higher FPS in the 21 day group relative to the 1 day group, indicative of spontaneous recovery, was observed when the extinction session was conducted 1 or 72 hours, but not 10 min after conditioning.
These results could not be explained by failure to extinguish when extinction training was conducted at the 10 min interval, as these rats showed decreased FPS relative to a group exposed to the context without extinction training of the light. Furthermore, in experiments where all sessions were conducted in the same context, rats’ FPS 24 hours after extinction did not significantly differ as a function of conditioning-extinction interval. Therefore, despite an apparent trend in the direction of weaker extinction at shorter conditioning-extinction intervals, there was no reliable evidence for differential strength of extinction between these groups. Thus, evidence for the potential of fear to return after extinction is strongest at the longest conditioning-extinction interval used
(72 hours), still partially present at the 1 hour interval, and absent at the 10 min interval.
More recent findings suggest that those reported by Myers et al. (2006) may lack generality. Maren and Chang (2006) found that extinction training conducted in the same context as conditioning after a 15 min interval failed to produce long-term extinction, as measured with freezing in rats 24 or 48 hours later, while a 24 hour conditioning-extinction interval allowed subsequent expression of a long-term extinction memory. This seemed to depend upon differences in context freezing at the beginning of the extinction session, which was higher at the shorter conditioning- extinction interval. Using a weaker conditioning procedure and shifting contexts
29 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
between conditioning and extinction sessions, so as to minimise context fear, allowed long-term memory of extinction trained at the 15 min interval. Sensitising contextual fear through delivery of shock in a novel context 15 minutes before an extinction session conducted 24 hours after conditioning prevented long-term extinction memory.
Thus, Myers et al.’s (2006) finding that extinction was possible at a 10 min interval may have depended on low levels of context fear, as would be expected in their design, which employed a context shift between conditioning and extinction.
Studies of extinction after a short post-conditioning interval in humans also cast doubt on the generality of Myers et al.’s (2006) findings. Using a within-subjects ABB vs ABA design, Alvarez, Johnson, and Grillon (2007) found evidence for renewal of eye blink startle, skin conductance response, and subjective anxiety in the presence of a tone that had been paired with shock and then subjected to extinction training in a different virtual reality context within minutes of conditioning. Schiller et al. (2008) employed a procedure in which extinction exposures to a fractal pattern began 12 s after the last image-shock pairing. Unsignalled shock exposure 24 hours later caused reinstatement of the skin conductance response 48 hours after extinction which was dependent on the reinstating shock exposure occurring in the same context as that used for other procedures. In rats, Schiller et al. (2008) also observed equal magnitude of reinstatement of freezing whether extinction training was commenced within 15 minutes of the end of conditioning or delayed until 3 days later.
These reports suggest that Myers et al.’s (2006) findings may be restricted to very specific experimental parameters. Nevertheless, Myers et al.’s (2006) findings do suggest that under certain circumstances, certain variables can affect the behavioural consequences of extinction exposures, possibly by differentially affecting the recruitment of different learning mechanisms. This would imply the possibility that
30 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
each ‘component’ of extinction may be further discernable in terms of its neurobiological substrates and behavioural properties. Overexpectation, the decrement in fear CRs that can occur as a result of CS - US pairing if the US delivered is in some way lesser in magnitude than that predicted on the basis of the stimuli present, is a preparation that can isolate the contribution of a specific mechanism to decrements in fear. This is because the use of US-delivery during overexpectation training should exclude the development of any ‘CS – no-US’ memory and should also mitigate against the development of habituation to the CS or degradation of a US representation. Among the models of error-correction discussed so far, only those that assume the contribution of a pooled error term (e.g. Rescorla & Wagner, 1972) can account for overexpectation.
Thus, if such a mechanism contributes to extinction learning, independently of any “no-
US” learning that proceeds, overexpectation should share with extinction any properties specific to those components of extinction produced by this mechanism. For these reasons, it is of interest to review the study of overexpectation, including its parallels with extinction.
3. Overexpectation
In a typical overexpectation experiment, two CSs ‘A’ and ‘B’ are separately paired with a US. Following acquisition of a strong CR to each CS, they are presented in compound and further paired with the same US. Despite this continued CS - US pairing, conditioned responding to each CS on its own is reduced as a result of this compound conditioning. Equation 1 models this effect by assuming that in the initial conditioning of the CSs, the internal associations between each CS representation and that of the US grow sufficiently strong that the combined associative strengths of the two CSs exceeds that which can be conditioned by the US (VA + VB > λ). Thus, when the compound
31 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
stimulus is paired with the US, the parenthetical term of Equation 1 takes a negative value (λ − ΣV < 0), meaning that the pairing results in a decrement in subsequent fear
CRs.
Rescorla (1970) first observed this effect by conditioning rats to separately associate both a tone and a flashing light with a 0.5 s, 0.5 mA shock. Rats then received additional pairings of either a compound of the tone and flashing light, only the flashing light, or only the tone with the 0.5 s, 0.5 mA shock, or received no additional conditioning. Rats that had received the compound conditioning showed less suppression of food-reinforced bar-pressing to subsequent presentations of either the tone or light alone than rats in the other groups. This effect has been replicated in fear conditioning in rats using suppression of food-reinforced bar pressing (Kremer, 1978), lick suppression (Blaisdell, Denniston, & Miller, 2001) and freezing (McNally, Pigg, &
Wiedemann, 2004a) as measures of fear.
Overexpectation appears in other species and conditioning preparations, having been observed in eye-blink conditioning in rabbits, autoshaping in pigeons, appetitive magazine-approach conditioning in rats, and causal judgement learning in humans.
Khallad and Moore (1996), for example, conditioned pigeons to associate localised white images of a square or ‘X’ on a key-light with 3 s availability of food. In a second stage of conditioning compound presentation of square and X was reinforced with food.
If the magnitude of reinforcement of the compound was equal to that used during elemental conditioning, subsequent pecking of each image presented on its own was reduced relative to pre-compound conditioning rates. If, however, reinforcement was doubled during compound conditioning by following each compound with two 3 s food presentations, this overexpectation effect was not seen. This is consistent with the prediction of Equation 1 that increasing the λ value of the US during compound
32 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
conditioning can attenuate or abolish overexpectation. This demonstrates that the decrement after compound conditioning is not a general effect of any compound conditioning procedure, regardless of the events that follow the compound presentations, but is specific to a situation in which Equation 1 would predict a negative prediction error.
In appetitive conditioning in rats, overexpectation has been replicated in several published experiments (Lattal & Nakajima, 1998; Rescorla, 1999; 2006; 2007). Lattal and Nakajima (1998) reinforced a light, noise, and tone CS with delivery of a food pellet 88 times each, after which a compound of the light and one of the auditory CSs was paired with food an additional 24 times. On a subsequent test of each auditory CS, the one used in compound conditioning evoked less magazine responding than the other. This was true whether the other auditory CS had received no further conditioning after stage I or had received compound conditioning in combination with a stimulus exposed but not reinforced during stage I conditioning. The lower responding to the auditory CS conditioned in compound with an excitatory visual CS was also reflected in a significant decline in response rate between the end of stage I conditioning and test, a decline not seen to the control stimulus. These demonstrations confirm the findings of
Khallad and Moore (1996) in pigeons, and further support the contention that the overexpectation effect is a specific result of negative error correction rather than a generalised consequence of compound conditioning. Furthermore, Lattal and Nakajima
(1998) showed that when the reinforcement of the CSs during conditioning was made conditional on an instrumental lever-press response during CS presentation, this instrumental responding also showed an overexpectation effect. Rate of lever-pressing during presentations of an auditory CS that received stage II conditioning in the
33 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
presence of a visual CS previously paired with food was lower than in the presence of the control CS.
Rescorla (1999) demonstrated that overexpectation could be observed in appetitive conditioning when each CS was conditioned with one of two different reinforcers with similar motivational valence during Stage I conditioning. A sucrose solution and food pellet served as outcomes 1 and 2 (O1 and O2, counterbalanced). A visual CS was paired with O1 and 2 auditory CSs paired with O2 in stage I conditioning. In stage II conditioning a compound of the visual CS and one of the auditory CSs was paired with one of the outcomes, while the second auditory CS continued to receive pairings with
O2. Overexpectation of magazine entries elicited by the tone used in compound conditioning was observed. These data suggest that at least some component of the prediction that leads to overexpectation in appetitive conditioning is based on the general motivational properties of the reward, rather than sensory properties specific to the US.
Other conditioning preparations in which overexpectation has been observed include eye-blink conditioning in rabbit (Kehoe & White, 2004) and causal judgement learning in humans (Collins & Shanks, 2006). In the experiments reported by Kehoe and White
(2004), CSs were first individually paired with shock to rabbits’ eyelids. In a second stage of conditioning, 2 CSs were presented in compound and this compound continued to be reinforced with shock. This resulted in a decline in probability of the eye-blink
CR, relative to CR probability at the end of stage I conditioning, on test trials delivered during stage II conditioning in which these CSs were presented individually, regardless of whether or not these test trials were also reinforced with shock. This decline was also reflected in lower CR probability on these trials than in a control group in which CSs continued to be individually reinforced during stage II conditioning. In rabbits
34 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
conditioned with a control stimulus that was paired with shock alone in both stage I and
II, CR probability on test trials of the CSs used in compound conditioning was lower than CR probability to this control stimulus during stage II conditioning and during a subsequent non-reinforced test of all three CSs.
Collins and Shanks (2006) subjected human participants to a causal learning task in which they were exposed to ‘records’ of an imaginary experiment examining the effect of different ‘types’ of radiation on cell mutation. 8 different types of radiation (CSs) were associated with differing levels of mutation. In a second stage, 2 different CSs (A and B) previously paired with the same level of mutation were presented in compound and paired with the same level of mutation that they were individually associated with in stage I. 2 other CSs (E and F), both associated with another level of mutation, were also presented in compound during this stage, and this compound was associated with twice the level of mutation that each of these two CSs had previously been associated with. All 8 CSs used in stage I conditioning were then individually tested by asking participants to rate the strength of the mutation they were likely to cause. Estimated levels of mutagenic potency declined between the end of stage I and test for CSs A and
B, but not for any other stimuli.
These demonstrations of overexpectation suggest that it is not an artefact of specific conditioning preparations or species. It has been observed in rats, pigeons, rabbits, and humans, in appetitive, aversive, and motivationally non-valent conditioning preparations. Controls used in various experiments have shown that it cannot be readily explained by over-training of excitatory conditioning or any general decremental effect of compound reinforcement. These experiments provide strong support for the contention that when two CSs are presented in compound, their associative values summate, and any discrepancy between expectation and outcome produced by this
35 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
summation has consequences for future behavioural responses to either CS. Thus, overexpectation appears to provide support for elemental, rather than configural, models of stimulus processing, and particularly those models, such as the Rescorla-Wagner model, that propose that associative change is regulated by a common error term, rather than separate error terms for each CS.
According to the Rescorla and Wagner (1972) model, overexpectation and extinction share a common cause (negative prediction error). Empirically, they share a common consequence (reduction in the CR). This leads to the hypothesis that overexpectation and extinction share common mechanisms and that various behavioural phenomena associated with extinction should be seen with overexpectation. Given a multi- component view of extinction, however, it is unclear how complete such an overlap should be. For example, mechanisms allowing the formation of a ‘CS – no-US’ memory in extinction, or contributing to the degradation of a US representation would not appear to contribute to overexpectation, in which a US is present during trials that lead to a reduced CR. Furthermore, prediction error resulting from pooled associative strength may differentially contribute to different components of extinction (e.g. erasure vs. inhibition), leading to greater commonalities between some extinction-associated phenomena than others in overexpectation. Finally, there may be reason to believe that any contribution of response-habituation to extinction would be less likely to occur during overexpectation, in which repeated presentation of a US would have a dishabituating influence.
Theoretically important behavioural parallels with extinction have also been detected by recent studies of spontaneous recovery and context-mediated renewal after overexpectation in appetitive conditioning. Rescorla (2006) conditioned rats to associate two visual and two auditory stimuli with delivery of a food pellet. Following 10 days of
36 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
this conditioning, 3 daily sessions were conducted in which one of the auditory CSs and one of the visual CSs were presented in compound, again paired with delivery of a food pellet. The ability of both stimuli of one modality to elicit entry into the magazine where food had been delivered was tested 1 hour after the final session of compound condition, whereas both stimuli of the second modality were tested a week later. An overexpectation effect, whereby the stimulus that had been used in compound conditioning elicited less CRs than its control stimulus, was observed during the 1 hour test, but not during the 1 week test, suggesting spontaneous recovery of the CR initially reduced by overexpectation.
Using a similar appetitive conditioning procedure, Rescorla (2007) demonstrated context-dependency of expression of overexpectation. In the first experiment reported, a light was paired with food in both of two contexts. In one of these contexts, a noise was also paired with food and in the other, a clicker was paired with food. In a second stage, one of the auditory CSs was conditioned in compound with the light in the context where it had not appeared in the first stage of conditioning. Testing of the appetitive CR was then conducted for both auditory stimuli in both contexts. Presentation of the auditory stimulus that was used in the compound conditioning elicited less magazine entries than the control CS when tested in the same context used for compound conditioning, but not when tested in the context in which it had been originally conditioned (ABA vs. ABB renewal). Rescorla (2007) also showed a similar effect when the context in which a stimulus was conditioned alone and in compound was the same, and testing occurred either in the same or another context (AAB vs. AAA renewal). This is unlikely to be due to the different associative histories of the contexts, as renewal was still seen when each context was matched for its history of association with both elemental and compound conditioning. Furthermore, if conditioning of all
37 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
three stimuli occurred in one context, followed by reinforced pairings of the light in compound with one auditory stimulus in a second context and in compound with the other auditory CS in a third context, the CR elicited by each auditory CS on test was lower in its own compound conditioning context than in the context where it had not been presented (ABC vs. ABB renewal).
These demonstrations that overexpectation, at least in appetitive conditioning, is expressed specifically in the context in which compound conditioning occurred, along with the demonstration of spontaneous recovery from overexpectation, are theoretically important for several reasons. Firstly, they are at odds with the implication of the
Rescorla-Wagner rule that overexpectation is simply partial erasure of the CS - US association. Instead, as with extinction, at least some component of the loss in the CR is due to a failure to express some of a remaining CS - US association. The inhibition of this association is context-specific and vulnerable to the passage of time.
However, not all accounts of the inhibition of a CS - US association after extinction can provide a satisfactory account of overexpectation. Among accounts that propose the formation of a new association during extinction which subsequently inhibits or competes with the CS - US association for expression, those that specifically describe a
‘CS – no-US’ association (Pearce & Hall, 1980; Bouton, 1993) appear unable to explain how a similar memory could be learned with continued CS – US pairings. This does not exclude the possibility that overexpectation involves the establishment of a new association that, on test, can compete with the CS – US association established during initial conditioning for expression. The occurrence of a US during stage II conditioning that is weaker than expected may cause the formation of a new ‘CS – weak-US’ association, despite the absolute strength of the US not differing between stage I and stage II. On test, contextual and other variables may favour retrieval of either the ‘CS –
38 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
strong-US’ association established during stage I conditioning or the ‘CS – weak-US’ association established during stage II. Thus, Bouton’s (1993) theory may account for these findings if extended to allow alterations in the magnitude of the expected US relative to the magnitude of the actual US to cause the formation of distinct US representations. Nevertheless, even if such processes do contribute to overexpectation, it does argue against the necessity for a ‘CS – no-US’ memory for producing decrements in fear that show many of the same characteristics as those produced by extinction, such as renewal and spontaneous recovery.
The model proposed by Wagner (1981) could account for an inhibitory memory component common to both extinction and overexpectation. If simultaneous presentation of multiple CSs results in a sufficiently high proportion of US representation elements entering the A2 state of working memory prior to actual US delivery, then the proportion of US representation elements available to be excited to
A1 by the US itself will be relatively low. In such a situation, inhibitory learning resulting from simultaneous activation of CS elements in A1 and US elements in A2 may exceed any excitatory learning resulting from the US elements that are activated to
A1, resulting in a net outcome of inhibitory CS - US learning despite pairing of the CS and US.
In addition to loss of the CS - US association or development of an inhibitory CS -
US association, compound conditioning of two stimuli may also cause the learning of a within-compound ‘CS-CS’ association. According to a class of theories described as
‘comparator models’, it is this association, rather than any negative error correction, that is critical for the learning of overexpectation. In this model, as formalised by Stout and
Miller (2007), acquisition of association between any concurrently present stimuli, whether or not they are motivationally significant, proceeds independently of the
39 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
presence and associative strength of other stimuli, according to an error-correction rule.
Any notion of inhibitory associative strength is rejected by this model. Performance of the response to a CS ‘X’ presented alone, after pairings of a compound of CSs X and J with a US ‘O’, can, however, be modulated by concurrent, indirect activation of the US representation through a CS-CS association according to the equation:
RX = VXO – kf(VXJVJO) (3)
In other words, the conditioned response (R) to X reflects the associative strength between X and O minus a function of the product of the strengths of X’s association with J and J’s association with O.
This theory does not share the prediction, made by models of overexpectation based on negative prediction error during compound conditioning, that overexpectation is essentially the same form of learning as extinction. Instead, the memory process that produces overexpectation is described as being essentially the same as that which produces blocking and over-shadowing. The proposed mechanisms for expression of these phenomena are depicted graphically in Figure 1. Two stimuli, X and J, presented in compound at any point in conditioning will develop associations with each other at the same time as developing independent associations with any outcome, O, presented.
When X is presented alone later, it triggers recall of internal representations of both O and J in proportion to the respective strengths of each association. The internal representation of J then indirectly excites its own association with O. The directly and indirectly excited O representations are then processed by a ‘comparator process’ which subtracts from the directly-excited representation’s ability to trigger a response in proportion to the strength of the indirectly-excited representation.
40 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Figure 1. A pictorial depiction of the comparator hypothesis obtained from Stout, C. C. and Miller, R. R. (2007). Sometimes-competing retrieval (SOCR): A formalization of the comparator hypothesis. Psychological Review, 114, 759-783.
41 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
The competition between the directly- and indirectly-activated US representations can only occur if the representations share a similar spatio-temporal relationship to X and J respectively. Thus, altering the temporal and/or spatial relationship between J and
O after compound conditioning should cause recovery of responding to X presented alone. Furthermore, simply weakening the J-O association after overexpectation, as well as over-shadowing and blocking, should cause a recovery of conditioned responding to
X. Blaisdell, et al. (2001) reported evidence for both types of recovery from overexpectation in a conditioning preparation in which fear of two distinct auditory CSs was indexed by their ability to suppress licking of a water spout in water-deprived rats.
If extinction training was conducted with one CS after it received overexpectation training, recovery of the longer-lasting suppression of licking to the other CS was seen.
This effect was stimulus specific, because extinction of another shock-associated stimulus that had not been presented during compound conditioning did not prevent overexpectation from being observed.
Changing the temporal relationship between one CS and the US after compound conditioning also produced recovery from overexpectation. If a CS used in compound conditioning, where the US co-terminated with the compound, received additional conditioning in which US presentation was shifted to 5 s after CS offset, length of suppression induced by the other CS on test recovered to levels seen in controls that received no compound conditioning. If the CS receiving trace conditioning was one not used in compound conditioning, or if the continued conditioning of one stimulus after compound conditioning was conducted at the same CS - US interval as that used in compound-US pairings, overexpectation was observed.
As the ability of a US to condition a CR is reduced at increasing CS - US intervals, the effect of changing from co-terminating CS and US presentation to trace
42 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
conditioning may result from the same mechanism as the effect of post-compound- conditioning extinction. If this is the case, these effects can also be accounted for by models that do not appeal to a comparator process influencing response, but instead allow that within-compound associations can mediate retrospective revaluation of an absent CS upon further conditioning or extinction of the other CS presented in the compound. Van Hamme and Wasserman (1994) allow that a CS - US association of a non-present CS previously associated with another CS through compound presentation can change in the opposite direction of any associative change in the second CS resulting from its presentation. Dickinson and Burke (1996) extend Wagner’s (1981) model to allow the formation of excitatory associations between two stimulus representations simultaneously excited to the A2 memory state. Either of these models can account for recovery of a CR to a non-present CS after the association of the other
CS used in compound conditioning is weakened through further presentation of the second CS either alone or in a weaker association with the US.
However, neither of these models can clearly explain an additional finding reported by Blaisdell et al. (2001). When initial individual conditioning of the CSs and compound conditioning both involved a trace conditioning procedure in which the US was presented 5 s after CS offset, recovery from overexpectation to one stimulus was seen if the other stimulus was presented in further post-compound conditioning trials in which it co-terminated with the US. As this change in the CS - US interval should strengthen the association, Van Hamme and Wasserman’s (1994) model would predict enhanced overexpectation to the other CS. The prediction that would be made by
Dickinson and Burke’s (1996) model in this situation is less clear as it may differ given different assumptions about the strength and timing of activation of US elements to A2.
The comparator model is unique in its ability to clearly predict this effect.
43 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
Consistent with the comparator model’s claim that overexpectation shares common mechanisms with over-shadowing and blocking, rather than extinction, these effects have been observed to show similar vulnerability to recovery after post-compound- conditioning manipulations of one CS. Blaisdell, Denniston, and Miller (1999) report that shifts from delay to trace or trace to delay conditioning between conditioning of one stimulus in compound with another and one of the stimuli alone reverses the overshadowing of the other stimulus when tested for its ability to induce lick- suppression. This recovery of responding to one stimulus in a compound also occurs after extinction of the other stimulus in overshadowing (Matzel, Schachtman, & Miller,
1985) and blocking (Blaisdell, Gunther, & Miller, 1999). In order to distinguish between contributions of negative error correction and comparator processes in overexpectation, further investigation of similarities and dissociations between overexpectation and extinction and between overexpectation and blocking at the behavioural and neurobiological levels are of interest.
4. Chapter 1 conclusion.
The demonstrations of various fear reactions in numerous species to stimuli that do not possess inherently aversive properties, after their pairing with an aversive US, suggest a widespread ability to learn associations between contiguously presented stimuli across the animal kingdom. Extinction of conditioned reactions suggest that the internal system that supports associative fear learning is sufficiently versatile to allow animals to adjust their CR based on changes in the CS - US relationship. Indeed, post- extinction phenomena such as renewal, reinstatement, and spontaneous recovery, suggest that this system allows internal CS representations to hold multiple associations, and has mechanisms for selecting which of these associations CS presentation may
44 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
excite in different situations. However, phenomena such as blocking and overexpectation are widely seen as evidence that simple temporal contiguity of a CS and US is insufficient to produce associative learning. Similarly, protection from extinction suggests that simple non-reinforcement of a feared CS is insufficient to produce decrements in CRs. Instead, such phenomena are consistent with the necessity of predictive error for producing associative change. Both reinforcement and non- reinforcement of a CS may produce increments, no change, or decrements in CRs, depending on the degree to which other stimuli present excite or inhibit prediction of reinforcement.
Overexpectation is a particularly interesting phenomenon because it involves decrements in a CR despite the continued reinforcement of a CS. This makes it a useful preparation for studying the specific content and consequences of negative prediction error more generally, because the decrement produced by the prediction error is not confounded by forms of learning specific to non-presentation of an expected US. Put differently, comparison of extinction and overexpectation may allow more understanding of the relative importance of the quantitative prediction error, theoretically common to both overexpectation and extinction, versus the qualitative prediction error (presence versus absence of an expected US) specific to extinction, in the generation of response loss. However, this view of overexpectation is challenged by a set of theories, generally referred to as ‘comparator models’, that attribute the response loss seen in overexpectation to interference with retrieval of the CS - US memory on test, caused by a within-compound CS-CS association. According to this view, the study of overexpectation still has theoretical value, not for understanding the mechanisms of extinction, but for understanding the means by which the multiple associations learned during conditioning interact to modulate later response
45 Chapter 1. Behavioural characteristics of the learning and loss of conditioned fear.
performance. The following chapter will review research examining the neurobiological substrates of the acquisition and loss of conditioned fear and discuss its relevance to the behavioural data and theoretical material discussed in this chapter.
46 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Chapter 2
The neurobiology of the acquisition,
expression, and loss of conditioned fear
The purpose of this chapter is to review current knowledge and theory of the neurobiological substrates of the acquisition and expression of the forms of learning discussed in the previous chapter (fear conditioning, blocking, extinction, and overexpectation). This discussion is, necessarily, selective, focusing particularly on data that has implications for the theories of extinction discussed in the last chapter.
However, it is first necessary to give an overview of the biological mechanisms of fear association formation. As the amygdala is the key structure in the mammalian brain for the formation and expression of fear associations, empirical data and theoretical models regarding its role in this learning will be reviewed.
Theoretical models of the amygdala’s role in fear conditioning propose that it is the site of convergence for temporally contiguous CS and US signals. However, as discussed in the previous chapter, temporal contiguity of a CS and US is insufficient for learning. Thus, biological mechanisms involved in calculation of prediction error must also be understood. Therefore, empirical data regarding common biological mechanisms for limiting the acquisition of fear and allowing blocking, overexpectation, and extinction will be reviewed. The role of opioid receptor signalling, including within the midbrain periaqueductal gray matter (PAG), suggest that it is the critical neural mechanism for signalling prediction error, and potential pathways through which amygdala and PAG activity may interact to limit increments and decrements in association will be discussed. 47 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Following this, the literature regarding the biological mechanisms of extinction memory formation and expression will be reviewed, with a view to identifying possible neural substrates of erasure, ‘new learning’, and habituation components of extinction.
While amygdala and PAG activity are involved in both acquisition and extinction of conditioned fear, extinction engages several additional structures not necessarily involved in acquisition. Findings in recent years have led to a widespread acceptance of neuroanatomical models of extinction that propose a central role for the ventromedial prefrontal cortex (vmPFC) in the expression of an inhibitory extinction memory, particularly through modulation of GABAergic neurons in the amygdala. In addition, the hippocampus is widely believed to mediate the contextual dependence of extinction memories through its interactions with the vmPFC and amygdala.
However, the almost exclusive focus on extinction in investigations of the neurobiological investigations of fear loss has meant that it remains largely unknown whether these findings apply generally to loss of fear CRs, including overexpectation, or are highly specific to extinction. Thus, it is unclear to what extent each putative biological mechanism of extinction requires detection of predictive error. Indeed, only one published study (McNally et al., 2004a) has sought to investigate the neurobiological substrates of overexpectation. The results of this study, which showed that injection of the opioid receptor antagonist naloxone before compound conditioning prevented overexpectation, are consistent with overexpectation sharing similar mechanisms with extinction (McNally and Westbrook, 2003). However, this finding does not distinguish overexpectation from blocking, which can also be prevented by naloxone. Thus, while a parsimonious account of these results could be that opioid receptor activation mediates US prediction in overexpectation, blocking, and extinction, naloxone may also be disrupting the formation of within-compound associations in
48 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
blocking and overexpectation, thus preventing the action of comparator processes on
CR expression. Thus, in concluding this chapter, the theoretical importance of investigating the mechanisms of expression of overexpectation, including the role of
GABAergic signalling, will be discussed.
1. The role of the amygdala in acquisition of conditioned fear.
Extensive evidence from lesion, electrophysiological, fMRI, drug infusion, and genetic manipulation studies has suggested that the amygdala, a temporal lobe structure in the mammalian brain, is critical for formation of CS - US associations in Pavlovian fear conditioning. Among the earliest such studies was that of Kellicutt and
Schwartzbaum (1963), who administered electrolytic lesions of the entire amygdala to rats. Rats then bar-pressed for food during 15 daily sessions. Once per session, they were exposed to a 3 min tone which co-terminated with a 1 mA, .5 s shock. All 5 rats in a sham-lesioned control group learned a criterion-level suppression of bar-pressing to the tone by the end of this conditioning, 4 of them learning to this level within 3 sessions. Not a single lesioned rat met the experimenters’ criterion on any of the 15 sessions. Increasing the shock did not abolish this profound deficit in learning of the conditioned response (CR).
When lesions are administered before conditioning, as in Kellicutt and
Schwartzbaum’s (1963) study, it is unclear whether they are primarily disrupting learning or performance of the fear CR. Kellicutt and Schwartzbaum’s (1963) lesions also destroyed most of the amygdala, which is now known to contain 13 separate and functionally unique nuclei, some with further functionally distinct subdivisions. In addition, these lesions mostly destroyed parts of surrounding structures and, being electrolytic, would have destroyed fibres of passage projecting from neurons with soma
49 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
in other possibly distant structures. So the possibility that the learning deficit observed was mediated by extra-amygdala structures could not be excluded.
All these concerns, however, have been addressed by further studies, which have confirmed the necessary role of the amygdala in Pavlovian fear conditioning, and clarified the contribution of particular nuclei of the amygdala to acquisition and expression of different CRs. The nuclei of the amygdala most relevant to fear conditioning generally are the lateral (LA), basal (BA; also known as basolateral), basomedial (also known as accessory basal or AB) and central (CeA) nuclei (LeDoux,
2000). The LA, BA, and (sometimes) AB are often described collectively as the basolateral complex (BLA). In the case of acquisition of conditioned freezing to a tone in rats, the list of critical nuclei can be narrowed further to the LA and CeA. Nader,
Majidishad, Amaropanth, and LeDoux (2001) made electrolytic lesions to either the
LA, CeA, BA, AB, medial (MeA), BA and AB together, or entire amygdala before conditioning rats to fear a tone using a .5 s, .5 mA shock. CeA, LA or total amygdala lesions reduced freezing to the tone measured 24 hours after conditioning to a level not significantly different than that shown by controls that were exposed to unpaired presentations of the tone and shock. BA, AB, BA + AB, and MeA-lesioned rats froze similarly to un-operated and sham operated controls.
Studies using temporary functional inactivation can avoid the confound inherent in pre-conditioning permanent lesions, which do not allow distinctions to be drawn between effects on acquisition vs. performance of conditioned fear, and allow investigation of specific phases of acquisition, consolidation, and expression of a CS -
US association. Infusion of the GABAA receptor agonist muscimol into the BLA prior to tone-shock pairings leads to profound deficits in freezing to the tone and to the context during later, drug-free tests (Muller, Corodimas, Fridel, & LeDoux, 1997). The
50 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
deficits detected by Muller et al. (1997) could not be explained as resulting from state- dependent memory, as infusion of muscimol before both conditioning and testing had the same effect as infusion only prior to conditioning (or only prior to testing, showing that this area of the amygdala plays a role in the expression, as well as acquisition, of fear associations). Muscimol did not induce a long-term disruption of freezing behaviour in rats conditioned under saline infusion, tested under muscimol, and then re- tested drug-free.
Intra-BLA infusions of muscimol immediately after, rather than before, conditioning fail to induce such a deficit (Wilensky, Schafe, and LeDoux, 1999). While this suggests that neural function in the BLA is necessary for the acquisition, but not consolidation, of conditional fear, this conclusion is disputed by a study using a sodium channel blocker, rather than a GABAA receptor agonist, as an inactivating agent. Sacchetti,
Lorenzini, Baldi, Tassoni, and Bucherelli (1999) infused the sodium channel blocker tetrodotoxin (TTX) into the BLA at various post-conditioning intervals. Infusions up to
48 hours post-conditioning caused deficits in freezing to both tone and context stimuli
48-72 hours post-infusion. While this difference in findings between Sacchetti et al.
(1999) and Wilensky et al. (1999) may result from the different drugs used, it may also be due to larger numbers of rats per group being used by Sacchetti et al. (1999; 8-11 rats per group) than by Wilensky et al. (1999; 6 rats per group), allowing greater power to detect differences.
The LA functions as an early point of convergence for transmission of sensory information regarding auditory CS and aversive somatosensory US stimuli, making it well-placed to play a primary role in the formation of associations between these stimuli. The rat LA receives projections from the medial geniculate body (MG) of the thalamus, which in turn receives auditory information from the inferior colliculus
51 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
(LeDoux, Farb, & Ruggiero, 1990). These projections are necessary for the acquisition of conditioned freezing, as disconnection of the MG from the amygdala, through combined unilateral, electrolytic MG lesions and contralateral, ibotenic acid lesions of the amygdala disrupt learning of freezing to a tone paired with shock (Iwata, LeDoux,
Meeley, Arneric, & Reis, 1986). LA neurons respond to foot-shocks in anaesthetised rats and, in the dorsal subnucleus of the LA (LAd), a high proportion of neurons that respond to auditory ‘click’ stimuli also respond to foot-shock (Romanski, Clugnet,
Bordi, & LeDoux, 1993).
The importance of LA neurons to the learning of conditioned fear to auditory stimuli is further supported by the detection of plasticity in field potentials in rats (Rogan,
Staubli, & LeDoux, 1997), mice (Rogan, Leon, Perez, & Kandel, 2005), and cats
(Collins & Pare, 2000; Pare & Collins, 2000) and unit activity of these neurons in rats
(Quirk, Repa, & LeDoux, 1995; Repa et al., 2001) and cats (Collins & Pare, 2000; Pare
& Collins, 2000) in response to acoustic stimuli paired with shock. These increased responses were specific to a CS previously paired with shock. Where control CSs, which did not signal shock (CS-), were used in the same animals receiving the CS+
(Collins & Pare, 2000; Rogan et al., 2005) the change in the LA neurons’ responses to the CS- were generally in the opposite direction to the change in response to the CS+.
Where control measurements were analysed in a separate group of animals that received unpaired presentations of tone and shock (Quirk et al., 1995; Repa et al., 2001), no changes in tone-evoked LA responses were observed.
Repa et al. (2001) found that the earliest detectable changes in unit activity in the
LAd preceded the earliest detectable changes in conditioned freezing and suppression, consistent with the proposition that this plasticity is a necessary stage in behavioural learning. While Repa et al. (2001) found that LAd neurons with the shortest-latency
52 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
clicker responses, located at the dorsal tip of the LAd, showed only transient plasticity,
Maren (2000) noted potentiated spike firing of short response latency LAd neurons even after extensive overtraining (75 trials) of a noise-shock pairing. Neurons with slower responses, generally located in the more ventral parts of LAd, showed long-lasting changes in responsiveness to the clicker (Repa et al., 2001). Conditioning-related plasticity in the ventral sub-division of the LA (LAv) developed later in conditioning than in the LAd (Quirk et al., 1995). These data led Repa et al. (2001) to suggest that the short-latency, dorsal neurons received direct CS and US input and their activation, in turn, causes plasticity in more ventral neurons that store long-term fear memory.
1.1. Theories regarding the role of the amygdala in storage of fear memory.
This view of the LA as a critical site for the storage of fear memories is central to the neuroanatomical model advanced by LeDoux (2000) and Davis (e.g. Davis, Falls,
Campeau, & Kim, 1993; Davis & Whalen, 2001), in which CS - US associations stored in the LA allow subsequent CS presentations to trigger output of fear responses through the LA’s projections to the CeA. CeA activity is responsible for a central state of fear, with individual projections of the CeA controlling specific CR components. This
‘orthodox’ model has not gone unchallenged, and has been modified in more recent years. Early criticism of this model came from Cahill and McGaugh (1998) who argued that the amygdala’s main function in fear memory storage was to modulate memory storage in other brain structures, and to control performance of fear responses. Much of the evidence they review in support of this model comes from studies of inhibitory avoidance, in which fear, conditioned with context-US pairings, is measured by testing the animal’s avoidance of the contextual CS. This paradigm thus differs in two ways from that employed by most studies that form the basis of the ‘orthodox’ model, which
53 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
measure fear through responses to an unavoidable, unimodal CS. Indeed, Kellicutt and
Schwartzbaum (1963) found that rats with lesions of the entire amygdala, which abolished learning of conditioned suppression, could still learn not to bar-press if each bar press was punished with shock, suggesting that avoidance learning is less amygdala- dependent than learning of Pavlovian fear responses.
Another assumption of the ‘orthodox’ model of amygdala function in fear conditioning that has been challenged is the serial organisation of information processing. This model proposes that the CS - US association is formed and stored in the BLA and it is only after the formation of this association that CS information is able to excite the CeA mediated state of fear. In this model, the CeA is a passive output structure, dependent on transmission of fear-related information from the BLA before it can itself activate fear responses, rather than a site of learning-related activity. However,
Wilensky, Schafe, Kristensen, and LeDoux (2006) found that infusing muscimol into the CeA before (but not immediately after) tone-shock pairings produced a profound deficit in freezing to the tone 24 hours later, relative to an artificial cerebrospinal fluid
(ACSF) infused control group. This was not a state-dependent effect, as infusions either only before test or before both conditioning and test produced similar deficits. Neither did the deficit appear to be attributable to changes in shock sensitivity, as shock-induced movement was unaffected by the infusions relative to the ACSF-infused control group.
The volume of the infusion used did not appear to affect responsiveness of LA neurons’ field potentials to stimulation, so it is unlikely that this freezing deficit was caused by action of CeA-infused muscimol leaking into the LA.
Furthermore, rats with neurotoxic, preconditioning BLA lesions have been shown to acquire context and tone freezing similar to that shown by sham-lesioned rats after 75- trial over-training (Zimmerman, Rabinak, McLachlan, & Maren, 2007). The learning
54 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
and expression of this fear are both CeA dependent, as reversible CeA inactivation with muscimol before either conditioning or test prevented this freezing. These results suggest that CeA-dependent mechanisms could take over fear learning in the absence of an intact BLA, given enough conditioning. Findings such as these have led to modifications of the model of the amygdala’s role in fear conditioning to allow a greater role for acquisition related plasticity in the CeA as well as the LA. A comparison of the
‘orthodox’ and modified models is presented in Figure 2.
In the original model, direct LA-CeA projections allowed CS - US associations stored in the LA to trigger the CeA output pathways that led to production of a fear CR.
However, the medial subnucleus of the CeA (CeM), which projects to the brainstem structures responsible for fear CR production, does not receive direct inputs from the
LA. Instead the LA projects only indirectly to the CeM through other CeA subnuclei, the BA, and the GABAergic intercalated neurons (ITC). Projections from several subcortical structures may carry both CS and US information directly to the CeM and/or other CeA subnuclei which project to the CeM. Thus the updated model (Pare, Quirk, &
LeDoux, 2004; Wilensky et al., 2006) allows plasticity mediated by LA inhibition of
ITC neurons disinhibiting CeM neurons and/or direct association of CS and US information in the CeA to accompany CS - US association in the LA in fear memory formation.
1.2. The neurochemistry of fear conditioning
In both the BLA and CeA, activation of NMDA receptors is essential for the acquisition of Pavlovian fear memories. In the BLA, this was first demonstrated by
Miserendino, Sananes, Melia and Davis (1990) who infused the NMDA receptor antagonists AP5 or AP7 before presenting light-shock pairings. Both drug treatments
55 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Figure 2. A comparison of the ‘orthodox’ model (a) and a revised model (b) of amygdala connections involved in fear conditioning. In the figure depicting the revised model, ‘ICM’ refers to the intercalated cell mass (ITC). Image obtained from Wilensky,
A. E., Schafe, G. E., Kristensen, M. P., and LeDoux, J. E. (2006). Rethinking the fear circuit: The central nucleus of the amygdala is required for the acquisition, consolidation, and expression of Pavlovian fear conditioning. The Journal of
Neuroscience, 26, 12387-12396.
56 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
blocked acquisition of FPS as tested 1 week post-conditioning. Infusions 5 days after conditioning or immediately before testing had no effect. This effect was unlikely to have been caused by drug-induced changes in shock-sensitivity, as AP5-infused rats showed equivalent shock-induced activity to vehicle-infused controls at a range of shock levels.
AP5 similarly blocked acquisition, but not expression, of FPS to an auditory CS
(Campeau, Miserendino, & Davis, 1992). Maren, Aharanov, Stote, and Fanselow (1996) demonstrated similar deficits in acquisition of context freezing after BLA infusions of
AP5 before, but not immediately after, conditioning and confirmed that this was not due to state-dependent effects. Unlike Miserendino et al. (1990) or Campeau et al. (1992),
Maren et al. (1996) found that AP5 also disrupted expression of context fear. As Maren et al. (1996) used a dose of AP5 twice as high as that used in the other two aforementioned studies, and measured fear with freezing rather than FPS, it is unclear whether this discrepancy represents a dissociation between the effects of AP5 on expression of context vs. light or tone fear, freezing vs. FPS, or a general dose- dependent effect on fear expression.
Goosens and Maren (2003) extended these findings by showing that AP5 infusions into either the BLA or CeA before either tone-shock or context-shock pairings blocked acquisition of conditional freezing. Interestingly, upon further conditioning, rats that received CeA infusions of AP5 reacquired freezing to the tone or context faster than controls that had not received prior pairings, suggesting savings of the original conditioning. The fact that BLA-infused rats did not show this savings effect suggests that the BLA could learn the association in the absence of NMDA receptor function in the CeA, but that this learning could not be expressed without further conditioning. The
CeA, on the other hand, could not learn the association in the absence of NMDA
57 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
receptor function in the BLA, at least given the number of pairings (5) used in the original conditioning.
The activation of amygdala NMDA receptors, including those containing NR2B subunits appears to be necessary for both short and long-term storage of conditioned fear memory. Rodrigues, Schafe, and LeDoux (2001) found that administration of ifenprodil, a selective antagonist of the NR2B subunit of the NMDA receptor, 15 min before conditioning, dose-dependently attenuated both tone and context fear acquisition.
Effects on freezing were seen whether test was conducted 1 hour or 24 hours after conditioning, at doses that did not disrupt expression of conditioned freezing. Similar effects were seen when the drug was infused into the LA 30 min before conditioning.
Bauer, Schafe, and LeDoux (2002) showed that intra-LA infusion of AP5, the antagonist effects of which are not specific for any particular NMDA receptor subunits,
10-15 min before conditioniong with tone-shock pairings, similarly impaired conditioned freezing at 1, 3, 6,and 24 hours post-conditioning.
NMDA receptors contain calcium-permeable channels which, when opened, increase intracellular calcium ion levels, triggering signalling cascades that allow both short- and long-term synaptic plasticity. Another type of calcium channel implicated in fear memory formation is the L-type voltage-gated calcium channel (LVGCC). Unlike the
NR2B-regulated NMDA receptor channel, opening of the LVGCC is only implicated in long-term memory formation. Intra-LA infusion of the LVGCC antagonist verapamil prior to tone-shock pairings in rats caused a dose-dependent reduction in tone-freezing at 24, but not 1, 3, or 6 hours post-conditioning (Bauer et al., 2002). Verapamil did not cause any change in post-shock freezing across a range of doses.
Long-term fear memory storage requires intracellular signalling cascades triggered by entry of calcium into cells through NMDA receptors and LVGCCs. Calcium influx
58 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
stimulates intracellular accumulation of cAMP (Wu et al., 1995). cAMP binds to the regulatory subunits of protein kinase A (PKA), leading to the release of catalytic subunits than then translocate to the nucleus and directly phosphorylate the cAMP- response element binding protein CREB. PKA-activity dependent phosphorylation of
CREB can also operate indirectly through interaction with a calcium-dependent cascade leading to the phosphorylation of mitogen-activated protein kinase (MAPK) (Impey et al., 1998). Calcium influx activates the protein Ras, allowing phosphorylation of the kinase MEK by MEK kinase, in turn allowing the phosphorylation of MAPK (Rosen,
Ginty, Weber, & Greenberg, 1994). In the rat amygdala, PKA activation regulates the
MAPK cascade through a phosphatidylinositol 3-kinase (PI-3 kinase) dependent mechanism (Lin et al., 2001). CREB phosphorylation promotes CREB’s association with a second protein, CREB-binding protein. This in turn allows the activation of
CREB-regulated genes (Frank & Greenberg, 1994). Transcription of CREB-regulated genes allows synthesis of new proteins that lead to long-term changes in neural morphology necessary for long-term memory storage.
The importance of these intracellular signalling cascades is supported by observations of correlation between activation of several stages of these cascades with
Pavlovian fear conditioning and is also demonstrated by the impairment in long-term fear memory by disruption of various stages of these cascades. PI-3 kinase is transiently activated in the BLA at 10, 30, and 40 (but not 5 or 60) min after light-shock pairings.
Inhibition of this activation with intra-BLA infusions of wortmannin before conditioning impairs fear memory (as indexed by potentiated startle) 24 hours (but not 1 hour) after conditioning (Lin et al., 2001). Following PI-3 kinase activation,
ERK/MAPK is transiently activated in the rat LA at 60 min (but not 15, 30, or 180 min) after paired (but not unpaired) tone-shock presentations. Infusion of the ERK/MAPK
59 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
inhibitor U0126 into the BLA before pairings dose-dependently impaired freezing to the tone at 6 or 24 hours, but not 1or 3 hours, post-conditioning (Schafe et al., 2000).
U0126 had no effect on post-shock freezing and U0126-infused rats could be retrained one week later to normal levels of freezing.
Furthermore, intra-LA infusion of Rp-cAMPS, an inhibitor of PKA activity, or anisomycin, a protein synthesis inhibitor in rats immediately after (but not 6 hours after) a tone-shock pairing also dose-dependently impaired freezing to the tone 24 hours, but not 4 hours, after conditioning (Schafe & LeDoux, 2000). Again, these effects did not appear to result from a permanent drug-induced inability to learn tone-shock associations and they also did not appear to result from state-dependent memory.
Similar results have been found when rats have been infused with the mRNA synthesis inhibitor actinomycin-D into the BLA before noise-shock pairings, with deficits in freezing detected to both noise and context 24 hours after conditioning (Bailey, Kim,
Sun, Thompson, & Helmstetter, 1999).
In addition to the BLA, protein synthesis is also necessary in the CeA for consolidation of conditioned freezing to a tone. Infusion of anisomycin into the CeA immediately after conditioning causes deficits in freezing at 24 hours, but not 4 hours, after conditioning (Wilensky et al., 2006). Finally, mice with a mutation of the CREB gene that causes them to lack the α and δ isoforms of CREB show impaired context freezing 1 or 24 hours (but not 30 min) and impaired tone freezing at 2 and 24 hours
(but not 30 min or 1 hour) after pairing with shock (Bourtchuladze et al., 1994.) These mutant mice flinched, jumped, and vocalised at similar minimum shock intensities as wild-type controls, suggesting equivalent shock-sensitivity. It must be noted, however, that in this study, the gene deletion was not tissue-specific, and so it cannot be
60 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
determined whether the behavioural effects were due to changes in the amygdala or in other brain structures.
The nature of the plasticity that follows CREB-mediated transcription is further explored in several studies of gene and protein expression following Pavlovian fear conditioning. Ressler, Paschall, Zhou, and Davis (2002) killed rats 0, 1, 4, or 8 hours after light-shock pairings. Immediate early genes (c-Fos and zif-268) were significantly induced in the amygdala, striatum, hippocampus, and cortex 0-1 hrs after conditioning.
Induction levels for these genes were back to baseline by 2 hours. Genes encoding structural proteins (eg. NF-1) also showed similarly early peaks in induction. NMDA receptor stabilisation protein and α-actinin expression peaked 1-4 hours after conditioning. RC3 neurogranin and gephyrin mRNA decreased during this time.
Gephyrin is involved in glycine and GABA receptor clustering. Another group of rats were exposed to light-shock pairings and killed 2 hours later. In BLA, α-actinin was significantly induced. α-actinin is involved in mediating clustering of NMDA receptors and other glutamate receptors. Gephyrin was significantly reduced in the BLA. Rats conditioned with odour-shock pairings showed very similar patterns of changes in induction.
These changes in gene expression lead to the hypothesis that BLA neurons show changes in glutamate and GABA receptor expression after fear conditioning. These hypotheses are confirmed by several studies. Yeh, Mao, Lin and Gean (2006) observed increased surface expression of GluR1 subunits of the AMPA receptor in the LA and
BA 2 and 24 hours after light-shock pairings relative to an unpaired control group, suggesting that the AMPA receptor is up-regulated in the storage of conditioned fear.
This up-regulation occurred without any significant overall change in GluR1 mRNA or protein levels, suggesting the increased surface expression was based on pre-existing
61 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
receptor proteins in the cytoplasm. This increase was also abolished by pre-conditioning intra-BLA infusions of AP5, or by sufficient pre-exposure to the CS to prevent learning of significant FPS to the light (latent inhibition). Yeh et al. (2006) also showed that intra-BLA infusion of a proteasome inhibitor, which promotes enhanced surface expression of GluR1, enhanced fear conditioning as measured by FPS 24 hours after conditioning. Using a virus-injection technique that allowed detection of GluR1 insertion into the synapse, Rumpel, LeDoux, Zador, and Malinow (2005) observed increased synapse-specific incorporation of GluR1 subunit in the LA 3 hours after tone- shock pairings in rats, relative to unpaired controls, in around 1/3 of LA neurons.
Another viral amplicon vector that prevented incorporation of new GluR1 into synapses reduced freezing 3 and 24 hours after conditioning, relative to rats injected with a control vector, when infused into the LA 14-20 hours before conditioning.
The findings reported by Ressler et al. (2002) regarding the GABAA receptor clustering protein gephyrin have also been replicated and expanded on. Chhatwal,
Myers, Ressler, and Davis (2005) found that gephyrin was significantly down-regulated in the rat BLA 2 hours after light-shock pairings compared with light- or shock-alone groups. Binding of H3 flunitrazepam, an agonist at the benzodiazepine binding site of the GABAA receptor, was significantly lower 6 (but not 2) hours after conditioning,
3 suggesting a decrease in surface expression of GABAA receptors. In mice, binding of H flunitrazepam has been observed to significantly decline in BA and CeA 2.5 hours after tone-shock pairings, relative to home cage controls (Heldt & Ressler, 2007). This reduction in GABAergic activity in amygdala neurons is a likely substrate of their increased excitability after Pavlovian fear conditioning.
GABAA receptors are a heterogeneous population of receptors, by virtue of their variegated subunit composition. While GABAA receptors are composed of 5 subunits,
62 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
there are at least 21 identified subunit types (α1−6, β1−4, γ1−4, δ, ε, θ, π, and ρ1−3;
Fritschy, Schweizer, Brunig, & Luscher, 2003). Receptors containing the α1, α2, α3, or
α5 subunits in combination with any β subunits and the γ2 subunit are most common, and contain benzodiazepine receptor sites (Mohler, Fritschy, & Rudolph, 2002).
Benzodiazepine receptors are expressed throughout the rat amygdala, and are most abundant in the LA, the anterior and posterior cortical nuclei, and the caudal portion of the posterior BA (Niehoff & Kumar, 1983).
A study using point-mutation in mice that renders specific subunits of GABAA receptors insensitive to diazepam found that inhibitory signalling in the BLA is mediated largely by α1 and α2 subunit-containing receptors, while in the CeA, this signalling is mediated almost entirely by receptors containing the α2 subunit
(Marowsky, Fritschy, & Vogt, 2004). However, Heldt and Ressler (2007) noted strong expression of mRNA for α3 subunits in LA and BA. In the CeA, α3 subunit labelling was weaker, and restricted to the CeM. α5 subunit labelling was much weaker throughout the three amygdala nuclei examined. Heldt and Ressler (2007) reported that changes in GABAA receptor expression after fear conditioning were specific to certain receptor subtypes. In mice exposed to tone-shock pairings and killed 2.5 hours after the end of conditioning, reduced α1 subunit mRNA labelling was seen in LA and BA relative to home cage controls. Despite its very low levels of expression, labelling of mRNA for α5 subunits was decreased in the CeA relative to home cage controls.
However, no significant changes in mRNA for α2, α3, or γ2 subunits were observed.
Altered expression of the receptor is not the only mechanism by which changes in
GABAergic inhibition may contribute to the consolidation of conditioned fear memories. Changes in the levels of GABA released by GABAergic interneurons in the
BLA may also be involved in conditioned fear. The GABAergic LA interneurons, 63 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
which mediate feed-forward and feed-back inhibition of glutamatergic projection neurons, are well-situated to undergo fear conditioning-related plasticity. In rats, these neurons receive convergent afferent projections from sensory cortical and thalamic areas, stimulation of which results in inhibitory post-synaptic potentials in principal projection neurons (Szinyei, Heinbockel, Montagne, & Pape, 2000). Szinyei,
Narayanan, and Pape (2007) showed that, in mice, these interneurons show a calcium- permeable AMPA receptor-mediated long-term potentiation after theta-burst tetanic stimulation of thalamic afferents. Fear conditioning (but not pseudo-conditioning) caused a significant decrease in the ability of such stimulation to induce this potentiation 24 hours, but not 2 weeks, after tone-shock pairings, despite no concurrent influence on baseline GABAergic transmission.
Pape and Stork (2003) detected a significant down-regulation of expression of the gene for the 65 kD isoform (GAD65), but not the 67 kD isoform (GAD67) of glutamate decarboxylase, an enzyme involved in GABA synthesis, 24 hours, but not 2 weeks, after fear conditioning in mice. In apparent contradiction to these findings, Heldt and Ressler
(2007) reported reduced mRNA expression for GAD67, but not GAD65, in LA of mice killed 2.5 hours after conditioning. This discrepancy may be due to the difference between these two experiments in the interval between conditioning and sacrifice of the mice. However, both experiments hint at the possibility of reduced GABA synthesis in
LA after fear conditioning. Together these results suggest that fear conditioning-related plasticity in GABAergic interneurons leads to reduced availability of GABA, which in turn may contribute to changes in excitability of projection neurons observed after fear conditioning.
64 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
2. The neuroanatomy of the expression of conditioned fear.
Following fear conditioning-induced plasticity, the amygdala structures involved in acquisition of conditioned fear are also involved in expression of that fear to subsequent presentations of the CS. In humans, magnetoencephalography shows amygdala activity preceding the predicted timing of US onset to a visual stimulus previously paired with an aversive noise burst, activation not seen to a similar, but non-conditioned stimulus
(Moses et al., 2007). Infusion of muscimol into the rat BLA prior to exposure to a context or tone previously paired with shock causes deficits in freezing (Helmstetter &
Bellgowan, 1994; Muller et al., 1997). Consistent with the theories that increased
AMPA receptor expression and/or increased probability of CS-elicited glutamate release in the BLA underlies conditioned fear memory, infusion of the AMPA/kainate receptor antagonist CNQX into the BLA prevents expression of FPS to a light or noise previously paired with shock in a dose-dependent manner (Kim, Campeau, Falls, &
Davis, 1993).
Zinebi et al. (2003) found that 48 hours after noise-shock pairings, paired pulse facilitation in LA neurons was reduced in fear-conditioned rats relative to naive rats and rats receiving unpaired presentations of the noise and shock. This finding suggests increased probability of transmitter release from glutamatergic, thalamic input neurons in fear conditioned rats. This is further supported by Venton, Robinson, Kennedy, and
Maren (2006). When exposing rats to a context or noise CS the day following completion of two days of CS-shock pairings, they detected an immediate, rapid increase in glutamate levels in the BLA with rapid-sampling micro-dialysis. Such changes were not seen in controls that had not experienced shock.
This effect on transmitter release may be mediated by LVGCCs. Protein expression of the α1C subunit of the LVGCC is increased in the rat amygdala following fear
65 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
conditioning. Systemic nimodipine, an LVGCC antagonist, blocks expression of FPS, but not baseline startle, in a dose-dependent manner. Nimodipine increases PPF in LAd neurons evoked by stimulation of the thalamo-amygdala pathway in brain slices of fear conditioned, but not naïve or unpaired control rats (Shinnick-Gallagher, McKernan,
Xie, & Zinebi, 2003). Together, these results suggest that increased LVGCC function in the LAd following fear conditioning supports increased probability of transmitter release following stimulation of pathways that carry CS information to the amygdala.
While the available evidence points to increased probability of glutamate release in the amygdala during expression of conditioned fear, evidence regarding changes in
GABA transmission is less clear. A microdialysis study in mice (Stork, Ji, & Obata,
2002) found a dramatic reduction, lasting at least 3 hours, in GABA levels in the left
BLA of mice when they were re-exposed to a tone previously paired with shock. This decrease was significantly larger than the more transient decrease observed in pseudo- conditioned mice, while no decrease was observed in unshocked controls. Stork et al.
(2002) collected dialysate samples at 20 min intervals, allowing only poor temporal resolution. Using rapid-sampling microdialysis that allowed monitoring of GABA levels at 14 s intervals in rats, Venton et al. (2006) observed changes in GABA levels during both conditioning and testing of fear that appear to contradict the results of Stork et al. (2002). The first of 6 (but not subsequent) noise-shock pairings elicited a rapid, but transient (lasting only min) increase in GABA in the BLA, as did exposure to a noise CS or feared context the following day. Comparison between these two apparently contradictory studies is made difficult by the different species used and vastly different temporal resolution of the microdialysis measures.
The BA seems particularly important for expression of fear in the intact brain. While rats with pre-conditioning electrolytic lesions of the BA are able to learn conditioned
66 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
freezing to a tone paired with shock, rats with post-conditioning BA lesions show a profound deficit in performance of this CR (Anglada-Figueroa & Quirk, 2005).
Similarly, rats with post-conditioning lesions of the CeA show failure to express FPS to a previously conditioned light (Hitchcock & Davis, 1986) or tone (Hitchcock & Davis,
1987). Thus, the LA and CeA are sufficient for acquisition and expression of fear in a brain with a damaged BA but, generally, the BA plays an important role in the expression of conditioned fear.
A recent paper using transgenic mice that allow long-lasting genetic tagging of c-Fos active neurons provides evidence supporting the hypothesis that neurons in the BLA activated during acquisition of fear memories are also involved in these memories’ expression. Reijmers, Perkins, Matsuo, and Mayford (2007) used expression of the gene
LAC, the expression of which could be maintained in the presence of doxycycline as an indicator of neuronal activity during acquisition of context-shock memory, and expression of the immediate-early gene Zif/Egr as an indicator of neuronal activity during retrieval of context-fear. They could thus identify neurons active during both acquisition and expression of conditioned fear by identifying those that expressed both
LAC and Zif/Egr. Of home cage, fear-conditioned, fear-conditioned/no-retrieval, and no-shock groups, only the fear conditioned groups that received the retrieval treatment showed overlapping expression of LAC and Zif/Egr that was significantly above chance levels. This group also had significantly more neurons showing overlapping expression than the home cage and fear-conditioned/no-retrieval groups. Reijmers et al. (2007) further observed that LAC-expressing neurons in the BLA were not GABAergic neurons, and so were probably projection neurons.
This is consistent with the model that fear conditioning-related plasticity allows reactivation, by CS presentation, of BLA neurons that encode fear learning events, and
67 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
which would otherwise not be sufficiently activated by CS-alone presentation to elicit fear responses. These neurons then project indirectly to CeA subnuclei where neurons are activated that, through their own projections, mediate behavioural expression of conditioned fear. Different projections from the CeA mediate different aspects of the coordinated physiological and behavioural response to a fear CS. For example, lesions of the lateral hypothalamus interfere with the conditioned arterial pressure response, but not freezing, to a tone previously paired with shock. On the other hand, lesion of the caudal PAG interferes with conditioned freezing, but not the conditioned arterial pressure response (LeDoux, Iwata, Cicchetti, & Reis, 1988).
In addition to the amygdala pathways discussed above, the medial prefrontal cortex
(mPFC) appears to play an important role in the expression of conditioned fear.
Corcoran and Quirk (2007) inactivated the prelimbic (PL) division of the mPFC, which projects robustly to the BA, LA, and CeA in rats (McDonald, Mascagni, & Guo, 1996), with TTX prior to pairing either a tone or context with shock, or prior to testing for freezing to the tone or context, or prior to exposure to a cat. TTX infused before conditioning reduced freezing during the conditioning session, but not on a subsequent drug-free test, showing that it did not prevent formation of tone or context fear. It also did not alter unconditioned freezing to the cat. However it did prevent conditional freezing to tone or context when infused before testing. TTX did not affect spontaneous locomotion.
Micro-stimulation of PL neurons 100-400ms after onset of a tone previously paired with shock has been shown to enhance freezing to the tone (Vidal-Gonzalez, Vidal-
Gonzalez, Rauch, & Quirk, 2006). Similar stimulation of the infralimbic cortex (IL), located ventral to the PL, appears to have the opposite effect, reducing expression of freezing to a CS (Milad & Quirk, 2002; Milad, Vidal-Gonzalez, & Quirk, 2004; Vidal-
68 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Gonzalez et al., 2006). Interestingly, though PL activity appears to be related to expression of fear, presentation of a feared tone CS in mice has been found to reduce overall unit firing in the PL (Garcia, Vouimba, Baudry, & Thompson, 1999). This decreased PL firing was not observed in mice with BLA lesions or when presentation of the CS was preceded by presentation of a light that had been trained as a conditioned inhibitor. This suggests that, while certain PL cells are involved in fear expression, a
‘tuning’ process may also take place to reduce extraneous activity, perhaps enhancing the signal: noise ratio of the fear-related activity.
This ‘tuning’ process may be mediated by dopaminergic activity. Presentation of a noise CS 2-3 hours after it was paired with shock elicits increased dopamine efflux in the rat mPFC, a change not observed in rats only exposed to the noise or to unpaired noise and shock presentations (Feenstra, Vogel, Botterblom, Joosten, & de Bruin,
2001). Pezze, Bast, and Feldon (2003) found that if the activity of this dopamine was altered by intra-mPFC infusion of either the indirect agonist D-amphetamine, or the
D1/D2 receptor antagonist cis-flupenthixol, prior to test of a tone 2 days after it was paired with shock, freezing to the tone was substantially reduced relative to a vehicle- infused group. These drugs showed similar effects if infused prior to both conditioning and test, suggesting that their disruption of the performance of the fear CR was not due to state-dependent memory. These drugs had no effect on freezing during conditioning sessions or test if infused only prior to conditioning.
3. Opioid receptor activation modulates detection of predictive error in both acquisition and extinction of conditioned fear.
Within the central nervous system of rats and humans, opioid receptor activation appears necessary for the limitation of acquisition of conditioned fear to a CS by both
69 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
its own pre-existing associative strength and that of other present stimuli. The opioid receptor antagonist naloxone enhances acquisition of conditioned freezing over multiple conditioning trials, as would be expected if it prevented the effects of US prediction from limiting further increments in associative strength. Fanselow (1981) presented rats with daily 4 min exposures to 2 different contexts. A 0.3 mA, 0.5 s shock was consistently presented 3 min after placement in one context, but not the other. Rats were injected with either 4 mg/kg naloxone or saline before these sessions. While naloxone did not increase freezing relative to saline-injected rats during adaptation to each context prior to conditioning, over the 8 days of conditioning it caused an increase in over-all freezing during all context exposures, and also increased the difference between freezing to the shocked and non-shocked contexts. On a test day conducted the day after the last day of conditioning, half the rats in each group had their drug treatment switched. Freezing to the shocked context, but not the non-shocked context, during this test was affected by which drug had been administered during conditioning, with naloxone-treated rats freezing more than saline-treated rats. However, freezing was unaffected by which drug was administered on the test day, showing that naloxone did not act to simply augment performance of the CR.
Using a similar conditioning procedure, Young and Fanselow (1992) conditioned rats with either no shock or 0.4, 0.6, or 0.8 mA shocks. While naloxone did not affect freezing in the non-shocked group, it increased asymptotic levels of freezing in the 0.4 and 0.6 mA groups to levels shown by both saline- and naloxone-treated rats in the 0.8 mA group, which appeared to freeze at ceiling levels. If, after 6 days of context conditioning using a 0.4 mA shock, rats drug treatment was switched during a second 6- day phase of conditioning, rats switched from naloxone to saline showed a decline in freezing, while rats switched from saline to naloxone showed an increase. This is
70 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
consistent with blockade of opioid receptors temporarily removing limits on increments on associative strength. Furthermore, Young and Fanselow (1992) did not observe reversal of latent inhibition of acquisition of fear to either a context or tone CS by naloxone, suggesting that its effects were not on CS processing.
McNally et al. (2004a) provided further replication of these effects of systemically injected naloxone on acquisition of context freezing, and extended these findings to a clicker CS. Naloxone injections prior to clicker-shock pairings led to slower extinction of freezing to the clicker during drug-free tests compared to saline-injected rats, consistent with super-conditioning of fear to the clicker in naloxone-treated rats.
McNally et al. (2004a) also investigated blocking of the clicker stimulus by pre- conditioning of the conditioning context. In rats injected with saline, this led to decreased freezing to the clicker alone relative to rats not exposed to context-shock pairings prior to conditioning of the clicker. However, in rats injected with naloxone, this blocking effect was not observed.
More recent experiments have shown that μ opioid receptors in the ventrolateral subdivision of the PAG (vlPAG) are particularly responsible for opioidergic contributions to the blocking effect in fear conditioning. These experiments have used the within-subjects blocking design where freezing to a stimulus ‘B’ conditioned in compound with a previously-conditioned stimulus A is compared to freezing to a stimulus ‘D’ conditioned in compound with an associatively neutral stimulus ‘C’.
Typically, less freezing is seen to stimulus B than stimulus D on test. Infusion of the μ opioid receptor antagonist CTAP into the vlPAG prior to compound conditioning has been found to abolish this blocking effect whether compound conditioning consists of 1
(Cole & McNally, 2007) or 4 (McNally & Cole, 2006) compound - shock pairings.
71 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
A role for the vlPAG in calculating predictive error is consistent with the implication of the Rescorla-Wagner rule that there must be some overlap or close interaction between systems responsible for calculating the associative strength of and/or responding to a CS and those responsible for processing the US. As reviewed previously, the PAG is responsible for the emission of the freezing response in rats. The vlPAG is also involved in processing the painful qualities of a US (Fanselow, 1998). In addition, there are several direct, and possibly other indirect, pathways that allow for interaction between the vlPAG and other areas of the brain involved in processing the
CS and US and emitting conditioned responses. The vlPAG and CeA share extensive reciprocal projections rats (Beitz, 1982; Rizvi, Ennis, Behbehani & Shipley, 1991). The vlPAG also receives projections from the mPFC, an area implicated in modulation of expression of conditioned fear and, in turn, may influence mPFC activity through indirect projections via the mediodorsal thalamus (Beitz, 1982).
Fanselow (1998) proposes that the mechanism by which CS-induced vlPAG opioid receptor activation modulates associative plasticity arising from US delivery is conditioned analgesia. According to this model, activation of CeM projections to vlPAG by the presence of stimuli that predict painful consequences controls an opioid receptor- dependent inhibition of pain-related input from the spinal cord. This pain-related input is necessary to support new fear conditioning-related plasticity in the amygdala. Thus, according to this model, acquisition of opioidergic conditioned analgesia, is key to the limitation of further fear acquisition by a CS’s pre-existing associative strength.
Conditioned analgesia, however, does not account for either the enhancement of second-order conditioning by naloxone (Cicala, Azorlosa, Estall, & Grant, 1990) or the prevention of fear extinction by antagonism of vlPAG opioid receptors (McNally Pigg,
& Weidemann, 2004b). In both cases, there is no painful US to inhibit the processing of,
72 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
yet there is still a presumed role for error-correction mechanisms, and still an effect of opioid receptor antagonism. Second order fear conditioning can occur when a neutral
CS is paired with a CS that has been established as a reliable predictor of a US, and which has become a fearful stimulus itself. Cicala et al. (1990) conditioned rats to associate a 10 s noise CS with a 2 s, 0.67 mA shock. After 10 such pairings, over 2 days, offset of a 10 s light was paired with onset of the noise CS five times, in a session that included no shock presentations. In a control group injected with saline before this session, this treatment was unable to establish the light as a second-order fear CS, as indexed by its ability to induce greater lick suppression than that seen in unpaired control groups on test. Injection of 2 mg/kg naloxone immediately prior to this session, however, allowed these CS-CS pairings to produce the second-order conditioning effect.
McNally et al. (2004b) infused naloxone into the vlPAG of rats before extinction training sessions of a tone that had been paired with shock one day prior to the commencement of extinction training. Eight 2 min tone presentations were presented during each of two extinction sessions. Naloxone infusion not only attenuated the reduction in freezing between the first and second extinction session, but also resulted in higher tone-freezing during a subsequent drug-free test, relative to saline-infused rats.
Thus, its effect was to attenuate extinction learning rather than performance or recall.
This effect was dose dependent and anatomically specific. Infusions into the nearby dorsolateral column of the PAG had no effect. To account for the effect of opioid receptor antagonism on acquisition, blocking, and extinction of conditioned fear,
McNally et al. (2004b) suggest that activation of these receptors allows a signal consistent with the -ΣV term in Equation 1 to be transmitted to the amygdala. This signal interacts with any US ‘λ’ signal, presumably in a structure or pathway upstream
73 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
of the amygdala, such that the amygdala processes a signal consistent with the parenthetical term of Equation 1 rather than simply a ‘raw’ US signal or, in the case of non-reinforcement, no signal. The detection of positive or negative prediction error results in plasticity involved in increments or decrements, respectively, in subsequent
CRs.
Consistent with this model, naloxone also prevents overexpectation of conditioned fear. McNally et al. (2004a) injected rats with of 3 mg/kg naloxone or saline before each of two sessions in which previously conditioned clicker and flashing light CSs were presented in compound and paired with shock. Rats injected with saline showed lower freezing to the clicker than controls that did not receive compound conditioning during a drug-free test the day following the second compound conditioning session.
Rats injected with naloxone did not show this overexpectation effect. Naloxone injection delivered in the absence of any compound conditioning did not affect subsequent freezing on test in the control group. This parallels McNally and
Westbrook’s (2003) finding that injection of 2.5 mg/kg naloxone before, but not after, extinction training of a clicker impaired extinction of freezing in a manner that could not be attributed to state-dependent memory.
4. The role of the amygdala in extinction
According to the view that the amygdala receives both positive and negative error signals, and initiates learning-related plasticity based on the valence and strength of these signals, the amygdala should be as important to extinction as it is to acquisition.
Indeed, as is the case with acquisition of fear conditioning, extinction requires activation of NMDA receptors in the amygdala. Infusion of AP5, but not CNQX, into the BLA immediately before extinction training of a light previously paired with shock
74 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
prevents the decrease in FPS between pre- and post-extinction tests that was seen in vehicle-infused groups (Falls, Miserendino, & Davis, 1992; Lin, Yeh, Lu, & Gean,
2003). This blockade of extinction by intra-BLA AP5 infusions has been replicated using freezing to tone and context CSs as the measure of fear (Lee & Kim, 1998).
Furthermore, Sotres-Bayon, Bush, and LeDoux (2007) found that both the within- session decline in freezing and recall of extinction during a drug-free test could be attenuated by intra-LA infusion of the NR2B subunit-specific NMDA receptor antagonist ifenprodil.
While involvement of NMDA receptor signalling in extinction seems common to a variety of circumstances, the roles of various downstream mechanisms appears to depend on the conditioning-extinction interval. Mao, Hsiao, and Gean (2006) found that extinction training of a light paired with shock decreased FPS relative to non-extinction controls whether the extinction session occurred 1 or 24 hours after extinction and that this decrease could be prevented by intra-LA infusion of AP5 or the PI-3 kinase inhibitor wortmannin 30 min before the extinction session. However, extinction conducted 1 hour after conditioning resulted in reversal of the conditioning-induced increase in surface expression of GluR1 expression in the BLA while, at the 24 hour interval, extinction training induced no change in GluR1 expression. At the short interval, the changes in subunit expression, like behaviour, could be blocked by AP5 and wortmannin infusions.
Further evidence for different mechanisms of extinction at different post- conditioning intervals comes from studies of the involvement of the MAPK signalling cascade in the amygdala. Intra-BLA infusion of inhibitors of MAPK or upstream activators of the MAPK signalling cascade, such as MEK or PI-3 kinase prior to extinction training conducted 24 hours after conditioning impairs extinction of FPS to a
75 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
light CS in rats (Lu, Walker, & Davis, 2001; Lin, Yeh, Lu, & Gean, 2003) or freezing to a tone in mice (Herry, Trifillief, Micheau, Luthi, & Mons, 2006). However, Herry et al.
(2006) found that when extinction training was conducted 5, rather than 24 hours, post- conditioning, intra-BLA infusion of MAPK inhibitor U0126 was without effect.
Extinction training at this interval also did not lead to alterations in pMAPK expression, while at the 24 hour interval, it led to an immediate, strong increase in pMAPK expression peaking at 60 min post-extinction and still significant 5 hours later. Thus, the
MAPK signalling cascade’s involvement in extinction may be specific to longer conditioning-extinction intervals.
Extinction of FPS also depends on protein synthesis, but possibly not mRNA synthesis. Lin, Yeh, Lu, & Gean (2003) found that intra-BLA infusion of anisomycin, but not actinomycin D, at doses that block acquisition of FPS, blocked its extinction.
However, the increased phosphorylation of CREB that is associated with these processes after fear conditioning was not seen in the BLA in extinction. Instead, the extinction session used by Lin, Yeh, Lu, & Gean (2003) reduced pCREB levels back to those of unpaired controls in rats decapitated 1 hour after extinction training. Both the post-conditioning increase and post-extinction decrease in pCREB was inhibited by PI-
3 kinase inhibitor.
In addition, the post-extinction decrease in pCREB is prevented by the calcineurin inhibitor FK-506. Calcineurin is a protein phosphatase, the levels of which increase in the BLA after an extinction session, peaking 20 min later and returning to pre-extinction levels 60 min after extinction. This calcineurin increase is blocked by pre-extinction infusion of anisomycin (Lin, Yeh, Lu, & Gean, 2003). Intravenous administration of 20 mg/kg calcineurin inhibitor cyclosporin A or intra-BLA infusion of FK-506 or cypermethrin blocks extinction training from inducing decrease in FPS.
76 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
The involvement of calcineurin activity is specific to extinction as calcineurin levels are not altered after conditioning relative to naive and unpaired controls and FK-506 does not block acquisition of FPS at the dose that blocked extinction (Lin, Yeh, Leu, et al., 2003). This involvement of calcineurin in extinction may appear at face value to be in contradiction to a role for the PI-3 kinase - MAPK cascade. Calcineurin dephosphorylates MAPK (Myers & Davis, 2007). Furthermore, phosphorylation of Akt, an activator of MAPK that is itself phosphorylated by PI-3 kinase, is significantly reduced in the BLA after extinction relative to non-extinguished controls matched for context exposure. FK-506, which does not alter Akt phosphorylation after conditioning, prevents this extinction training-induced reduction, suggesting that calcineurin activity inhibits this cascade (Lin, Yeh, Leu, et al., 2003). However, the involvement of both calcineurin and MAPK signalling in extinction may not be in contradiction if one considers the possibility that each process could be contributing to plasticity in different cell types. Calcineurin activity could be depotentiating activity and thereby weakening memory in one cell population (e.g. interneurons vs. principal cells, different types of interneurons, LA vs. BA cells, cells in different subdivisions of these nuclei, etc.) while the MAPK cascade is strengthening a new memory in another population of cells
(Myers & Davis, 2007).
The data reviewed above suggest that signalling in the amygdala may underlie several distinct extinction mechanisms including some that are consistent with an erasure view of extinction and some that are consistent with extinction involving new learning. In addition, it has been suggested that the contribution of amygdala cannabinoid receptors may underlie a habituation component of extinction. Systemic injection of the CB1 receptor antagonist SR141716A before extinction sessions has been found to prevent long-term extinction of conditioned freezing to tone (Marsicano
77 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Figure 3. Similarities and differences in intracellular signalling leading to consolidation of fear conditioning and extinction memories in the amygdala. Image obtained from
Lin, C., Yeh, S., Lu, H., & Gean, P. (2003). The similarities and diversities of signal pathways leading to consolidation of conditioning and consolidation of extinction of fear memory. The Journal of Neuroscience, 23, 8310-8317.
78 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
et al., 2002) and context (Suzuki et al., 2004) CSs in mice and of FPS to a light CS in rats (Chhatwal, Davis, Maguschak, & Ressler, 2005). Marsicano et al. (2002) also reported attenuation of within-session declines in freezing, though this was not observed by Suzuki et al. (2004). This effect is specific to extinction acquisition, as injection before a conditioning session had no effect on acquisition, and injection 10 min after an extinction session was also without effect (Marsicano et al., 2002). Enhancement of
CB1 activity through injection of the direct agonist WIN 55-212,2 augmented extinction of FPS in rats (Chhatwal, Davis, et al., 2005).
Kamprath et al. (2006) suggest that CB1 receptor activation specifically underlies a habituation component of extinction learning. CB1 deficient mutant mice or wild-type mice injected with 3 mg/kg SR141716A show impairments in habituation of unconditioned freezing to a non-conditioned tone 24 hours after sensitisation by unsignalled shock. These treatments attenuated both the within-session decline in freezing to the tone and recall of reduced freezing 5 days later. CB1 deficient mice were, however, capable of inhibitory learning to the tone produced by a backward conditioning procedure in which shock immediately preceded tone. Both CB1 deficient mice and wild type mice showed lower freezing to the tone than sensitisation controls when tested the following day, suggesting that the CB1 receptor is not necessary for learning of inhibitory memory to the tone. Kamprath et al. (2006) did not, however, test the effect of a CB1 antagonist in wild-type mice in this task. It can not, therefore, be determined whether this negative result is due to the lack of CB1 receptor involvement in this form of learning in wild-type mice or adaptations that occur in CB1 receptor- deficient mice that allow them to learn this inhibition through other pathways.
Studies measuring changes in field potential and unit activity in the lateral amygdala after extinction appear, on the surface, to be consistent with the view that extinction
79 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
involves reversal of conditioning-induced changes in activity in some, but not all, neurons potentiated by fear conditioning. Quirk et al. (1995) report that extinction training of a tone reversed the conditioning-induced increases in unit responses of most
LA cells that showed plasticity after tone-shock pairings. Rogan et al. (1997) report that
LA field potential responses to a tone, which were potentiated after its pairing with shock, returned to baseline levels after extinction training. In humans, amygdala activity, as measured by fMRI, shows a temporally graded decline during extinction training of a visual CS paired with shock, in contrast to the increased activation seen during conditioning (LaBar, Gatenby, Gore, LeDoux, & Phelps, 1998; Phelps, Delgado,
Nearing, & LeDoux, 2004).
However, synchrony in spontaneous firing of pairs of LA cells, which increased after tone-shock pairings, remained at post-conditioning levels after 30 extinction presentations of the tone (Quirk et al., 1995). This may reflect protection from extinction of some associative links in a layered network (Kehoe, 1988). Inputs carrying tone information to the LA may lose the ability to excite neurons after extinction, but associations between neurons within the LA formed during conditioning that allow synchronous activity remain intact. Furthermore, Repa et al. (2001) report that a population of cells located in the ventral portion of the rat LAd, with longer-latency responses to a tone CS than those located in the dorsal tip of LAd, maintained potentiated unit responses to a 20 s noise CS through 20 extinction trials. Thus different populations of LA neurons involved in CS responses may be differentially vulnerable to extinction-related depotentiation.
LA neurons that appear to lose their CS-responsive unit activity after extinction training may still show context-mediated renewal. Hobin, Goosens, and Maren (2003) conditioned rats to fear both a tone and white noise CS. Extinction training of each CS
80 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
was then carried out over three days (one session for each CS per day, beginning one day after conditioning). Each rat was then tested with presentations of each CS in each context over the following two days. As expected, freezing induced by CS presentations showed renewal, with each 2 s CS eliciting less freezing during the 1 min post-CS period in its own extinction context than in the extinction context of the other CS.
Accompanying this renewal, spike firing of LA neurons in response to the CSs was greater in the context inconsistent with the CS’s extinction than in the extinction context, particularly during the 40-50 ms and 60-70 ms post-CS onset time bins. This context-modulation of LA unit activity was specifically due to extinction training as it was not seen in rats subjected to similar procedures, but either not presented with shock during conditioning or not subjected to extinction training after conditioning.
This context-mediated return of unit responding after extinction suggests that declines in CS-elicited neural activity after extinction do not necessarily reflect simple reversal of acquisition-related plasticity. Instead, it suggests that activity of CS- responsive neurons is inhibited after extinction. This then raises the question of whether such inhibition is learned and expressed within the amygdala, or whether it recruits additional structures, including ones not necessarily involved in fear acquisition.
Current evidence suggests that intra-amygdala mechanisms may contribute, but that additional structures are also engaged in the consolidation and expression of learned fear inhibition.
Within the amygdala, GABAergic transmission may mediate inhibition of fear- related neural activity after extinction in several locations, including in the BLA and the
ITC mass. In the BLA, indicators of GABAA receptor expression generally show changes after extinction that are in the opposite direction to the changes seen after conditioning. Chhatwal, Myers, et al. (2005) observed that gephyrin mRNA levels in
81 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
the rat BLA were significantly increased relative to non-extinguished controls 2 hours after extinction training of a light conducted 48 hours after its pairing with shock.
Gephyrin protein levels were increased 6 hours after extinction training. Availability of benzodiazepine receptors, as measured by H3-flunitrazepam binding, was increased at both time points (Chhatwal, Myers, et al., 2005). In mice killed 1.5 hours after extinction training conducted 24 hours after conditioning, Heldt and Ressler (2007) observed increased GAD67 mRNA expression in BA relative to controls that received fear conditioning but remained in the home cage during the extinction session. α2 subunit mRNA labelling in the CeA and gephyrin mRNA labelling in BA also increased after extinction relative to a group exposed to the context but not to tone presentations, though neither of these two groups differed significantly in either of these measures from the home cage controls.
These results suggest that enhanced GABAergic activity in the amygdala is involved in the consolidation of extinction. Consistent with this suggestion, enhancement of
GABAA receptor activation after an extinction session has been found to enhance extinction. Akirav, Raizel, and Maroun (2006) infused a low dose of muscimol into the
BLA either before or after an extinction session involving 5 presentations of a 30 s tone that had been paired with shock the preceding day. Post-, but not pre-extinction infusions enhanced consolidation of extinction of freezing, as tested over the next two days in rats.
While this effect could be interpreted as disrupted reconsolidation of the fear memory after recall, rather than enhanced extinction, Harris and Westbrook’s (1998) study of the effects of FG7142 on extinction learning further supports the involvement of GABAA receptors in extinction learning. FG7142 is a non-subunit-specific partial inverse agonist at the benzodiazepine receptor which reduces the sensitivity of the
82 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
GABAA receptor to GABA. When injected subcutaneously (s.c.) at a dose of 10 mg/kg into rats 15 min before extinction training of a clicker that had been paired with shock the day before extinction, it reduced the within-session decrement in freezing to the tone and the memory for extinction the following day. This effect was dose dependent and was not due to state-dependent memory as injection of 10 mg/kg FG7142 before test did not produce recall of extinction in rats that received the drug prior to extinction.
In apparent contradiction to prevention of extinction learning by pre-extinction administration of a GABAA receptor inverse agonist (Harris & Westbrook, 1998) or the enhancement of extinction by post-extinction intra-BLA infusion of a GABAA receptor agonist (Akirav et al., 2006), Berlau and McGaugh (2006) report enhancement of extinction by a GABAA receptor antagonist. In their experiments, bicuculline was infused into the BLA immediately after both of two extinction exposures (1 per day) of a context that had been paired with shock the day before extinction training begun.
When freezing to the context was measured the day after the second extinction session, rats receiving these infusions froze significantly less than rats infused with saline or infused with bicuculline 3 hours after the extinction sessions. This apparent contradiction may reflect different mechanisms underlying extinction of fear to context vs. tone CSs.
The expression of extinction is also GABAA receptor-dependent. Harris and
Westbrook (1998) found that FG7142 increased freezing to an extinguished tone CS on when injected 15 min before test at a dose of 5 or 10, but not 2.5 mg/kg relative to vehicle-injected controls. AAB vs. AAA or ABC vs. ABB renewal occluded the effect of FG7142, suggesting that GABAA receptor activation mediates the expression of the context-specific component of extinction. GABAA receptors located in the BLA are at least partly responsible for GABAergic mediation of expression of extinction as intra-
83 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
BLA infusion of the GABAA receptor antagonist picrotoxin blocks expression of extinction without affecting expression of non-extinguished fear (Barad, Gean, & Lutz,
2006).
5. The prefrontal cortex in extinction.
Extinction learning can only partly be explained by neural activity within the amygdala itself. An important component of extinction learning appears to involve structures that are closely linked to the amygdala and come to control its activity after extinction training, including the mPFC and hippocampus. When the mPFC or, specifically, its ventral aspects (including IL and ventral PL) is lesioned electrolytically in rats prior to tone-shock pairings, extinction of freezing to non-reinforced tone presentations is delayed, though it can eventually be learnt. Morgan, Romanski, and
LeDoux (1993) lesioned rats before exposure to 4 tone-shock pairings over two days.
Beginning the following day, rats were exposed to a single daily presentation of the 20 s tone. Rats with electrolytic lesions took, on average, twice as many days to extinguish to a criterion of < 5 s freezing during the tone than sham-lesioned rats or un-operated controls. However, groups did not differ in the speed with which criterion levels of extinction to the context were achieved.
Lebron, Milad, and Quirk (2004) found that when 15 extinction trials were massed in one session beginning an hour after tone-shock pairings, lesioned rats showed the normal within-session decline in freezing. On test the following day, however, lesioned rats’ freezing was statistically indistinguishable from non-extinguished controls, and significantly higher than that shown by sham-lesioned rats, indicating failure to recall extinction. Lesions did not affect conditioned freezing in rats that did not receive extinction training. When spaced extinction trials (2 per day) were conducted after these
84 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
initial procedures, lesioned rats eventually showed evidence of extinction learning.
However, similar to the findings of Morgan et al. (1993), lesioned rats receiving only this spaced extinction took twice as many extinction sessions to show freezing levels significantly below those shown on the first extinction/test session than sham-lesioned rats.
Not all studies investigating the effect of mPFC lesions on extinction have replicated the effects reported by Morgan et al. (1993) or Lebron et al. (2004). Gewirtz, Falls, and
Davis (1997) found that electrolytic lesions of the rat mPFC administered between conditioning and extinction stages failed to attenuate extinction of FPS to a light CS or activity suppression to the conditioning context over 18 extinction/test sessions. Neither was extinction of activity suppression or FPS to a tone CS altered by pre-conditioning lesions. Garcia, Chang, and Maren (2006) also failed to find an effect of either pre- conditioning or post-extinction mPFC lesions on freezing during a 1 min period that followed a 2 s tone CS either during an extinction session or on test a week later.
There are several plausible explanations for the discrepancies between these findings. In the studies that found an effect of mPFC lesions on extinction, CS duration was either 20 s (Morgan et al., 1993) or 30 s (Lebron et al., 2004). In those that did not find an effect, CS duration was 2 s (Garcia et al., 2006) or 4 s (Gewirtz et al., 1997).
Thus, the discrepancy in findings may reflect a dissociation in mechanisms contributing to extinction of longer vs. shorter-duration CSs. Furthermore, the procedural parameters used by Gewirtz et al. (1997) induced very gradual extinction even in the control group.
If mPFC lesions only impair more rapidly learnt components of extinction, lesioned rats may not display significant retardation of extinction learning when compared to controls that only acquire slowly-learnt components of extinction. In the case of the study by
Garcia et al. (2006), freezing was measured after, rather than during CS presentation, an
85 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
important methodological difference with those studies finding an effect of lesions on extinction.
Another possible explanation for the lack of an effect of mPFC lesions on extinction reported in some studies is related to the degree to which the caudal IL was or was not spared by the lesions. Quirk, Russo, Barron, and Lebron (2000) administered pre- conditioning, electrolytic vmPFC lesions that either included or spared the caudal IL in rats. Acquisition of freezing and conditioned suppression of food-reinforced bar pressing to a 30 s tone paired with shock was not affected by lesion. Neither was the decline in both these measures seen during an extinction session that commenced 1 hour after the completion of conditioning. Recovery of fear, as indexed by both these measures during an extinction test conducted the next day, was significantly higher in rats with lesions including caudal IL than sham-lesioned rats. Freezing, but not suppression, in these rats was also significantly higher during this test than in rats with lesions sparing the caudal IL. Rats with lesions including caudal IL did show a faster decline in freezing and suppression during this session than non-extinguished controls, despite showing equivalent fear during the first trial of the session, suggesting that while recall of extinction was impaired, some extinction memory had been consolidated after the first extinction session. A specific role for the caudal IL in extinction recall may explain the discrepancy between the findings of Lebron et al. (2004) and Quirk et al. (2000) on the one hand and those of Gewirtz et al. (1997) and Garcia et al. (2006) on the other. Both Lebron et al. (2004) and Quirk et al. (2000) report lesions extending caudally as far as 1.7 mm anterior to bregma in the IL while Garcia et al. (2006) and
Gewirtz et al. (1997) report lesions only extending as far caudally as 2.15 and 2.2 mm anterior to bregma respectively.
86 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
To find an effect of pre-conditioning lesions on long-term extinction memory leaves open the question of whether the lesioned structure is involved in consolidation or recall of extinction (or both) or if the effects of lesions are even due to subtle effects on acquisition or consolidation of fear, which are only revealed after extinction. To answer these questions, Sierra-Mercado, Corcoran, Lebron-Milad, and Quirk (2006) inactivated the vmPFC with infusion of the sodium channel blocker tetrodotoxin either before tone- shock pairings, before or after extinction training of the tone conducted the following day, or before test on a third day. Infusion 30 min before conditioning produced no significant effects on freezing or suppression of food-reinforced bar-pressing to the tone during any stage of the experiment. Infusion 30 min before extinction training resulted in a non-significant decrease in freezing, and a significant decrease in suppression during the early extinction trials relative to vehicle-infused rats, consistent with an effect of PL inactivation on expression of fear (Corcoran & Quirk, 2007). The following day, rats receiving these infusions showed increased freezing, but not significantly increased suppression, relative to vehicle-infused rats during a drug free test. Infusion immediately after the extinction session did not result in the effect on freezing during test. This provides some evidence that vmPFC activity is involved in encoding extinction memory. The fact that this was not reflected in a blockade of a within-session decline in freezing suggests that the component of extinction encoded by vmPFC activity is a long-term component that is not immediately expressed in the time-frame of a 60 min session. The lack of effect of post-extinction infusion may suggest that vmPFC activity is unnecessary for consolidation of this component of extinction memory or that, if it is necessary for consolidation, this can occur at a later time point after the TTX is no longer active.
87 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Surprisingly, Sierra-Mercado et al. (2006) found that infusion of TTX 30 min before test on the third day reduced both freezing and suppression to the tone. This is in contradiction to the lesion data showing impaired recall of extinction (Morgan et al.,
1993; Quirk et al., 2000; Lebron et al., 2004). In the lesion experiments, substantial portions of the mPFC, including the PL, were damaged before conditioning. As the mPFC, particularly the PL, has been shown to play a role in acquisition and expression of conditioned freezing in the intact brain (Pezze et al., 2003; Laviolette, Lipski, &
Grace, 2005, Laviolette & Grace, 2006; Corcoran & Quirk, 2007), acquisition and expression of fear in permanently lesioned rats may have depended upon different neurobiological substrates than in non-lesioned rats. Fear may have been learned and expressed through these alternate pathways, but extinction learning dependent on the caudal IL (Quirk et al., 2000) could not proceed normally. In the non-lesioned rats used by Sierra-Mercado et al. (2006), expression of fear, including after extinction, was possibly controlled by PL. As the infusion sites reported by Sierra-Mercado et al.
(2006) included both IL and PL sites, and IL and PL have been reported to exert opposing influences on fear (Vidal-Gonzalez et al., 2006), the effects of TTX infusion on test are difficult to interpret.
5.1. Plasticity of IL neural activity is associated with extinction.
In any case, neural activity and plasticity in the vmPFC, particularly the IL, appears to be correlated with, and necessary for, extinction consolidation and recall. Milad and
Quirk (2002) reported that neurons in the rat IL did not show tone-evoked firing during pre-exposure to tone, pairing of the tone with shock, or extinction training of the tone conducted 1 hour after conditioning. They did, however, show tone-evoked firing 100-
400 ms after tone onset during a test the following day. The increase in firing to the tone
88 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
during this epoch correlated negatively with freezing during this test, consistent with a role for this neural activity in extinction recall. No such extinction-induced tone responses were observed in the PL or medial orbital cortex (an area rostral to IL).
If IL neurons were stimulated with electrodes 100-400 ms after onset of a tone presented the day following its pairing with shock, with no intervening extinction training, this stimulation not only simulated extinction by suppressing freezing, but enhanced extinction learning accruing from tone presentations. When tested for extinction memory the following day, rats that had received 20 or 100 Hz stimulation paired with tone presentations during the extinction session showed less freezing than non-stimulated rats (Milad & Quirk, 2002; Milad et al., 2004; Vidal-Gonzalez et al.,
2006). This stimulation was not effective if delivered 1 s before or after tone onset
(Milad et al., 2004). Therefore, artificial enhancement of IL activity can enhance extinction learning, but only if timed to simulate the natural IL activity seen during extinction recall. Milad and Quirk (2002) reported that IL stimulation could not function as a reward that motivated bar pressing. Therefore, its effects on extinction were probably not simply due to a reward state counteracting the behavioural effects of fear, but instead on the potentiation of processes naturally involved in extinction of conditioned fear. In support of the hypothesis that IL and PL neurons play opposing roles in conditioned fear, similarly timed stimulation of the PL had either no effect on extinction (Milad & Quirk, 2002) or impaired extinction of freezing as measured by a tone test the following day (Vidal-Gonzalez et al., 2006). Similar stimulation of the most dorsal areas of the mPFC (dorsal anterior cingulate and medial precentral cortex) had no effects on freezing or extinction (Vidal-Gonzalez et al., 2006).
The high-frequency stimulation of IL neurons in conjunction with tone presentation may have potentiated consolidation processes that naturally occur in these neurons
89 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
following extinction training, which later allow these neurons to signal extinction memory. In support of this hypothesis, increased firing of IL, but not PL neurons in the min and hours following extinction is predictive of successful recall of extinction the following day. Burgos-Robles, Vidal-Gonzalez, Santini, and Quirk (2007) exposed rats to tone-shock pairings and then to extinction training 2 hours later. Spontaneous activity in both the IL and PL was recorded for 5 min periods immediately before and 30, 60, and 120 min after the extinction session. When tested the following day, rats showed a bimodal distribution of freezing responses to the tone. Some showed recall of extinction, freezing less than 40% of the time that the tone was present, while others failed to recall extinction, freezing more than 60% of the time. Rats that recalled extinction showed significantly higher bursting activity in IL neurons at all post- extinction time points than rats that remained fearful of the tone. General firing rates in the IL were also higher in extinction-recalling rats at the 30 and 120 min post-extinction times. PL firing and bursting rates, however, did not distinguish between rats that recalled extinction and those that didn’t.
5.2. Role of NMDA receptors and intracellular signalling in mPFC in extinction.
The burst firing seen in the IL is NMDA receptor-dependent. Burgos-Robles et al.
(2007) report that systemic injection with 10, but not 5, mg/kg CPP selectively reduced the percentage of spikes occurring in bursts and within-burst spike frequency in vmPFC neurons 60 min after injection, without affecting average firing rates. Injection of the 10 mg/kg dose, but not the 5 mg/kg dose, prior to extinction training, or intra-vmPFC infusion before or immediately after (but not 2 or 4 hours after) extinction training also significantly impaired recall of extinction the following day. These results suggest that
NMDA receptor-mediated burst firing in the IL is critical to consolidation of a
90 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
component of extinction memory. This consolidation may allow potentiated IL neural responses to the CS on subsequent encounters, signalling recall of this component of extinction.
Consolidation of extinction memory involves activation of the ERK/MAPK signalling cascade and protein synthesis in the mPFC. When PD098059, an inhibitor of
MAPK activation, is infused into the mPFC immediately, but not 2 or 4 hours after, extinction training, rats fail to show evidence of extinction memory, as indexed by freezing to a tone CS either 1 or 10 days after the extinction session (Hugues, Deschaux,
& Garcia, 2004; Hugues, Chessel, Lena, Marsault, & Garcia, 2006). Extinction training was conducted 1 day after conditioning in these studies. Santini, Ge, Ren, de Ortiz, and
Quirk, (2004) report that intra-ventricular or intra-vmPFC infusion of anisomycin in rats
10 or 20 min, respectively, before extinction training of a tone CS (conducted 1 day after conditioning) produced amnesia for extinction during a test/extinction session the following day. The rate of within-session decline in tone-elicited freezing on test, relative to that shown by non-extinguished controls, yielded no evidence of savings in rats treated with intra-vmPFC anisomycin before extinction training, suggesting that this amnesia was total. This amnesia was not lifted by re-infusion of anisomycin before test, demonstrating that anisomycin did not induce state-dependent memory. Anisomycin had no effect if infused 4 hours after the extinction session or if infused before a session consisting of only 2, rather than 15 tone presentations, which produced no evidence of extinction learning the following day in vehicle-infused rats.
91 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
6. Inhibition of amygdala neurons by IL activity as a possible substrate for extinction.
The connections between the IL and the amygdala, and the manner in which IL neuronal activation modulates amygdala neuronal activity are consistent with the proposed role of the IL as a structure that expresses an inhibitory component of extinction by inhibiting output of fear-related signals from the amygdala. In rats, IL projections target the ventral subdivision of the LA, the AB, and the lateral capsular subdivision of the CeA. Caudal IL appears to have particularly strong projections to the intermediate and ventral subdivision of the CeA (McDonald et al., 1996). In cats, IL projections to the medial portion of the CeA, as well as the MeA have been observed
(Room, Russchen, Groenewegen, & Lohman, 1985) and in the Japanese monkey, projections to LA and BA have been observed (Chiba, Kayahara, & Nakano, 2001).
Rosenkranz and Grace (2002) found that mPFC stimulation caused burst firing in BLA interneurons while inhibiting projection neurons in anaesthetised rats. Rosenkranz,
Moore, and Grace (2003) also observed that mPFC stimulation hyperpolarised LA projection neurons and inhibited both spontaneous spike firing and spike firing elicited by an odour previously paired with shock in rats. However, neither of these studies distinguished between the IL and PL, limiting the conclusions that can be drawn from them.
mPFC stimulation appears to inhibit CeA projection neurons that are known to mediate emission of conditioned fear responses. Quirk, Likhtik, Pelletier, and Pare
(2003) observed responses of CeA neurons that project to brainstem structures to stimulation of the insula (in rats) or BLA (in cats). Animals also had stimulating electrodes targeted at the border between PL and IL. mPFC stimulation, which did not directly excite any CeA neurons observed in rats, completely blocked activation of
92 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
insula-responsive CeA neurons when delivered 10-20 ms before stimulation of the insula. In cats, mPFC stimulation, which only excited 1 of 69 CeA output cells observed, inhibited CeA unit responses to LA, BA, and AB stimulation in 74 % of CeA neurons observed, and also inhibited CeA field potential responses to BLA stimulation when mPFC was stimulated 10-60 ms before BLA stimulation.
mPFC-, and in particular, IL-mediated suppression of fear-related CeA activity is likely to be mediated through IL projections to the GABAergic intercalated (ITC) neurons. The ITC neurons are located in the lateral capsular subdivision of the CeA, to which the IL projects (McDonald et al., 1996), and project to the CeM. The ITC neurons also receive projections from LA. Pare et al. (2004) suggest that LA projection neurons inhibit ITC projections to the CeM by exciting other ITC neurons upstream of the ITC projection neurons, thus disinhibiting the CeA to allow expression of conditioned fear. IL projections to the ITC, according to this model, directly excite ITC projections to CeM, preventing input from the BA to the CeM from exciting expression of fear. In support of the hypothesis that IL inputs excite ITC neurons, Berretta,
Panatazopoulos, Caldera, Pantazopoulos, and Pare (2005) found that disinhibition of the
IL through infusion of the GABAA receptor antagonist picrotoxin led to increased c-Fos immunoreactivity in ipsilateral rostral ITC and contralateral caudal ITC relative to vehicle-infused controls in rats killed 135 min after infusion. No differences between picrotoxin- and vehicle-infused rats were seen in the lateral or medial subdivisions of the CeA, BA, or ipsilateral LA, though picrotoxin infusion significantly increased c-Fos immunoreactivity in the contralateral rostral LA.
93 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
7. The role of the hippocampus in contextual modulation of extinction.
The hippocampus appears to be involved in learning a context-specific component of extinction. Corcoran, Desmond, Frey, and Maren (2005) inactivated the dorsal hippocampus (DH) with infusion of muscimol prior to extinction training of a tone in rats one day after, and in a different context to that used for tone-shock pairings. When tested for freezing to the tone the following day, rats generally showed freezing levels consistent with at least partial learning of extinction (though the absence of a non- extinguished control means this assumption must be made with caution). However, while saline infused controls extinguished and tested in the non-conditioning context
(ABB) showed lower tone freezing than rats subjected to ABA, AAB, and ABC designs, rats infused with muscimol and subjected to the ABB design showed equivalently high tone freezing to muscimol and saline-infused rats subjected to these renewal designs. In other words, temporary inactivation of the DH during extinction training led rats to act as if they were showing renewal the following day in an extinction context. Thus, learning of the context-dependent component of extinction appears to rely on a functional DH.
Expression of a context-specific component of extinction also appears to involve hippocampal activity. Using an ABB design in humans, where the extinction/test context was distinct from the conditioning context, Milad et al. (2007) found that recall of extinction of a coloured light CS one day after conditioning and extinction training was associated with BOLD signals in both the vmPFC and right hippocampus. These
BOLD signals were correlated with the degree of extinction recall as indexed by percent change of skin conductance response to the CS between the maximum response during extinction training and the response at the beginning of the extinction recall test. vmPFC BOLD signal activation during this session also correlated significantly with
94 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
BOLD signals in both right hippocampus and amygdala, suggesting that vmPFC cooperates with both these structures in extinction recall. However, hippocampal activation during extinction recall was not seen in a study using an AAA design, where the same context was used for conditioning, extinction, and test (Phelps et al., 2004).
Milad et al. (2007) suggest that hippocampal involvement in extinction recall may be specific to contexts that are specifically associated with extinction.
The hippocampus also appears to be involved in signalling the renewal of conditioned fear in an ambiguous context. Using an ABB vs. ABC design, Corcoran and Maren (2001) found that, in rats receiving an infusion of saline into the DH, freezing to a tone previously paired with shock was significantly greater if tested in a non-extinction context than if tested in the extinction context. However, if the DH was inactivated with muscimol 20-25 min before the test, rats froze at equally low levels, similar to those shown by saline-infused rats tested in their extinction context, regardless of which context they were tested in. Muscimol infusion into the DH did not disrupt freezing to a non-extinguished tone, suggesting that hippocampal activity only came to modulate freezing after extinction had rendered it context-specific. Corcoran and Maren (2004) replicated this effect in an AAB vs AAA design, where conditioning and extinction are carried out in one context, and testing either the same or another context, but did not replicate it in an ABA vs ABB design. Therefore, when a test context is associated directly with the original conditioning experience extinguished fear can be renewed without a functional hippocampus. However, in an ambiguous context, hippocampal function is required for renewal.
Hippocampal control over contextual modulation of fear after extinction is likely to be mediated through its interactions with the vmPFC and BLA. A substantial proportion of neurons in the IL and ventral PL of rats exhibit excitatory responses to both BLA and
95 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
ventral hippocampal (VH) stimulation (Ishikawa & Nakamura, 2003). Simultaneous stimulation of VH and BLA greatly potentiated the excitatory responses of these vmPFC neurons. Stimulation of either structure 20-40 ms before stimulation of the other resulted in inhibition of the second structure’s excitation of these neurons. Lesions to connections between the VH and BLA do not alter these finding, suggesting that the convergence of VH and BLA signalling in vmPFC is not dependent on reciprocal connections between VH and BLA. Convergence of inputs carrying CS or fear expression-related information from the BLA (particularly the BA; McDonald, 1987) and contextual information from the hippocampus may induce plasticity that allows hippocampal inputs to influence mPFC-mediated expression of extinction. Consistent with this hypothesis, Hugues et al. (2006) found that after extinction training
(conducted in a context separate from that used for conditioning), mPFC field potential responses to VH stimulation were potentiated for at least 2 hours. Intra-mPFC infusion of PD098059, which attenuated later recall of extinction, also blocked this potentiation.
8. Summary of models of extinction and conclusion
Anatomical models of fear conditioning and extinction based on the data reviewed thus far hold that fear conditioning allows CS-related input to the LA from the medial geniculate nucleus of the thalamus (MG) to both disinhibit CeM neurons through LA projections to the ITC and excite CeM neurons through projections to the BA (Hobin et al., 2003; Pare et al., 2004; Maren & Quirk, 2004; Sotres-Bayon, Cain, & LeDoux,
2006). Context fear is also mediated through the amygdala by hippocampal projections to the BA (Sotres-Bayon et al., 2006). After extinction, CS information from the MG gains access to the vmPFC through the perirhinal cortex (Hobin et al., 2003), which then allows vmPFC activity to supress fear expression-related CeM activity through its
96 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
projections to the ITC and also possibly to inhibitory LA interneurons (Pare et al., 2004;
Sotres-Bayon et al., 2006). Learning about the occasion-setting properties of the extinction context is mediated by the hippocampus, which then gains the ability to, depending on the test context, either positively or negatively modulate mPFC activity and also to directly modulate LA activity (Hobin et al., 2003; Maren & Quirk, 2004;
Sotres-Bayon et al., 2006). This model, therefore, describes mechanisms that seem analogous to the theory of extinction associated with Bouton (e.g. 1993). The amygdala can be seen as the site of storage of a conditioning memory, the vmPFC the site of extinction memory storage, and the hippocampus the selector that disambiguates the meaning of a CS in a context-dependent manner. If selected for expression, the vmPFC- dependent extinction memory prevents expression of the competing memory by inhibiting amygdala neurons necessary for expressing the conditioning memory. The
GABAergic ITC neurons are a critical structure for this competition between the two
CS-linked memories. The involvement of GABAergic neurons in mediating a context- dependent inhibitory component of extinction memory is supported by the finding that interference with GABAergic transmission disrupts expression of extinction in a manner that does not summate with renewal (Harris & Westbrook, 1998).
However, some molecular data regarding extinction seems more consistent with erasure mechanisms than new learning mechanisms. These include reversal of conditioning-induced changes in AMPA (Mao et al., 2006) and GABAA (Chhatwal,
Myers et al., 2005) receptor expression in the BLA and the involvement of calcineurin signalling, which is involved in depotentiation (Lin, Yeh, Leu et al., 2003; Lin, Yeh,
Lu, & Gean, 2003). Furthermore, the apparent similarities between the roles of CB1 receptors in extinction (Marsicano et al., 2002; Suzuki et al., 2004; Chhatwal, Davis, et al., 2005; Barad et al., 2006) and habituation (Kamprath et al., 2006) suggest support
97 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Figure 4. A depiction of the proposed role of IL projections to the amygdala in modulating fear-related neural transmission. Glutamatergic LA projections to the ITC cells excite GABAergic inhibition of other downstream ITC cells which in turn disinhibits CeM output neurons which control fear responses. IL projections may excite
ITC neurons that inhibit CeM activity, counteracting the effect of LA projections.
Image modified from Pare, D., Quirk, G. J., & LeDoux, J. E. (2004). New vistas on amygdala networks in conditioned fear. Journal of Neurophysiology, 92, 1-9.
98 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
Figure 5. Depiction of a neural model of expression and modulation of extinction.
Information regarding a tone CS gains access to the LA, which can trigger a CeA- mediated freezing CR through various indirect projections, but which may also mediate some aspects of extinction. Tone information is also transmitted to the vmPFC, which can express tone-extinction memory through its projections to the ITC (‘ICM’ in this figure) and/or LA. The hippocampus mediates contextual memory, allowing context fear to be expressed through its projections to the BA (‘B’ in this figure), and mediates contextual control over expression of tone extinction through its projections to the vmPFC and/or LA. Image obtained from Sortres-Bayon, F., Cain, C. K., & LeDoux, J.
E. (2006). Brain mechanisms of fear extinction: Historical perspectives on the contribution of prefrontal cortex. Biological Psychology, 60, 329-336.
99 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
for the contention that extinction also involves habituation to the CS (McSweeney &
Swindell, 2002). Several studies strongly suggest that different mechanisms contribute to extinction depending on the time at which extinction training is conducted. These include the finding of differing vulnerability of extinction to reinstatement, renewal, and spontaneous recovery depending on the length of time elapsed between conditioning and extinction (Myers et al., 2006), the finding that changes in GluR1 expression were altered when extinction training was conducted 1, but not 24 hours after conditioning
(Mao et al., 2006), and the finding of differential involvement of MAPK activation in extinction training conducted 5 vs. 24 hours after conditioning (Herry et al., 2005).
These findings are consistent with the suggestion that there may exist multiple mechanisms of extinction, with the relative contribution of each mechanism to behavioural response decrements influenced by specific properties of the extinction training. The findings reported by Myers et al. (2006) and Mao et al. (2006) in particular suggest that erasure mechanisms may dominate extinction learnt soon after conditioning, but be less important for extinction learning at longer intervals. One question that remains unanswered is the degree to which each putative component of extinction involves error-correction mechanisms. To answer this question requires study of a preparation that can isolate the contribution of negative error-correction to losses in fear. According to the Rescorla-Wagner model, overexpectation, is an ideal preparation for such investigation. The experiments reported in this dissertation are an attempt to add to such investigation, particularly by exploring the role of GABAA receptors in the expression of overexpectation. These experiments examined whether expression of overexpectation of conditioned freezing was, like extinction (Harris & Westbrook,
1998), blocked by injection of FG7142. If the GABAergic contribution to the expression of extinction is learned through a negative error-correction-dependent
100 Chapter 2. The neurobiology of acquisition, expression, and loss of conditioned fear.
mechanism that also operates during overexpectation training, then expression of overexpectation should also be similarly vulnerable to FG7142. Specifically, if fear overexpectation, like extinction, involves upregulation of GABAA receptor subunits and benzodiazepine binding sites in the BLA and/or inhibition of CeM output neurons through activation of ITC projection neurons, FG7142 acting at GABAA receptors in the amygdala may reduce the inhibitory transmission supposedly involved in preventing expression of fear memory after extinction and, possibly, overexpectation.
101 Chapter 4. General Discussion
Chapter 3
Experimental reports
The experiments reported in this chapter explore fear overexpectation, the effect of the benzodiazepine receptor partial inverse agonist FG7142 on its expression, and the potential role of physical context in regulating its expression. Experiments also explored the effect of FG7142 on expression of fear not subjected to a decremental procedure, specifically in simple acquisition and blocking designs.
Experiment 1
Kremer (1978) exposed rats to 8 light-shock and 8 noise-shock pairings followed by
1, 4, 8, or 16 pairings of the light-noise compound with shock. The results showed that overexpectation, as measured by conditioned suppression, was most reliably observed with more rather than fewer Stage II pairings. Only one published report of overexpectation has used conditioned freezing as the measured response (McNally et al., 2004a). McNally et al. (2004a) employed a compound conditioning procedure over two days, with two pairings of a clicker-light compound per day.
Experiment 1 sought to provide a demonstration of fear overexpectation in freezing using only a single session of Stage II training. Experiment 1 also sought to establish whether overexpectation as measured by freezing was influenced by the number of
Stage II compound - shock pairings. The design is shown in Table 1. Rats were conditioned to fear both the tone and flashing light in separate sessions (Stage I). Stage
II conditioning involved either 2, 4, or 8 compound - shock pairings. Rats were tested the following day to presentations of the tone alone, and compared to a control group that received no compound - shock exposures. 102 Chapter 4. General Discussion
It was hypothesised that rats exposed to Stage II training would freeze less than controls. Furthermore, based on the results of Kremer (1978), it was hypothesised that the strength of this overexpectation effect would be positively correlated with the amount of Stage II training. Both error-correction and comparator accounts may make such a prediction if it is assumed that the learning that leads to overexpectation is not asymptotic after 2 compound - shock pairings. According to Equation 1, as long as the combined associative strengths of the two CSs exceeds the λ value of the US delivered, further compound conditioning will strengthen the overexpectation effect. According to a comparator account, the more the tone and flashing light are presented simultaneously, the stronger the association between them. This then allows subsequent tone-alone presentations to elicit indirect activation of the US representation.
Method
Subjects
Subjects were 55 experimentally naive male Wistar rats obtained from a commercial supplier (Gore Hill Research Laboratories, Sydney, Australia). After arrival, rats were housed in groups of 7 - 8 in plastic cages maintained on a 12/12 hour light/dark cycle.
They were allowed access to food and water ad libitum. The rats were handled for 1 min per day for 5 - 7 days prior to experimental procedures. The procedures were approved by the Animal Ethics Committee at the University of New South Wales and were conducted in accordance with the National Institute of Health (NIH) guidelines for the care and use of laboratory animals.
Apparatus
Behavioural procedures were conducted in a set of four identical chambers (195 mm height x 234 mm width x 204 mm length). The front and real walls as well as the hinged lid were made of clear perspex and the side walls were made of stainless steel. The floor
103 Chapter 4. General Discussion
consisted of stainless steel rods, 2 mm in diameter, spaced 13 mm apart (centre to centre). Each chamber stood 35 mm above a tray of paper pellet bedding (Fibrecycle,
Mudgeeraba, Australia). The chambers were cleaned with a damp paper towel and bedding changed between rats. These chambers were located individually within sound attenuating boxes, the inner walls of which were painted black. The boxes were illuminated with a red LED light. An extractor fan in the rear wall of each box was operating during all sessions.
The auditory CS was an 82 dB, 750 Hz tone (0.1s rise-fall time). The visual CS was a 4 Hz presentation of a white fluorescent light mounted on the ceiling of the sound- attenuating box, producing an illumination level of 75 candela/m2 within the chambers.
Both CSs were 30 s in duration and during Stage I and II co-terminated with a 1 s, 0.5 mA unscrambled AC 50 Hz shock from a contant-current generator that was delivered to the floor of each chamber. Digital video cameras were mounted on the rear wall of each box and connected to a digital multiplexer in an adjacent room that, in turn, was connected to a DVD recorder. The stimuli were controlled by computer (LabView,
National Instruments, Austin, TX).
Procedure
Stage I. During the mornings of Days 1 - 6 all rats were placed in the conditioning chambers for 20 min sessions during which they were exposed to 1 CS-shock pairing.
The CS (tone on Days 1, 3, & 5; flashing light on Days 2, 4, & 6) commenced 9 min 30 s after placement in the chambers. On the afternoons of Days 4 - 6 rats were placed in the chambers again for 10 min each, with no CS or US presentation, to reduce context fear.
Stage II. Rats in the control group received 30 s handling on the morning of Day 7.
Rats in the remaining groups were placed in the conditioning chambers and received
104 Chapter 4. General Discussion
either 2, 4, or 8 presentations of the compound visual + auditory CS co-terminating in shock. Onset of the first trial was 4 min 30 s after placement in the chambers, and inter- trial intervals were 4 min 30 s. Rats were removed from the chambers 5 min after the last shock. Thus, the session lasted 15, 25, or 45 min for groups Over - 2, Over – 4, and
Over – 8, respectively.
Test. On Day 8 all rats were placed in the chambers and exposed to four 30 s presentations of the tone. The first presentation began 4 min after placement in the chambers, and further presentations followed at an inter-stimulus interval (ISI) of 4 min
1 s.
Data Analysis
Rats were scored every 8 s for freezing during context exposures and every 2 s during CS presentations on each day. Freezing was defined as the absence of all movement other than that required for respiration (Fanselow, 1980). Freezing data were converted into percentages of observations scored as freezing during the observation period. In this and remaining experiments, any rat that showed < 30% freezing to the tone on both Days 3 and 5 in Experiments 1 and 2, or on Day 3 in Experiments 3, 4, and
9 was excluded from further procedures and/or data from these rats were not included in statistical analyses. This criterion was set after partial completion of Experiments 1-3, when it was observed that most rats that performed at this level during Stage I also showed very low freezing on test regardless of what procedures they were subjected to, suggesting that they had failed to acquire fear to the tone. In Experiment 1, 3 rats were excluded from stage II conditioning and test on the basis of this criterion. Thus, at test there were 13 rats per group.
Data from any rat that froze during 50% or more of the first 3 min of context exposure on the test day (pre-CS freezing) was also excluded from analyses. The
105 Chapter 4. General Discussion
rationale for this criterion is that freezing to the tone on test in these rats may reflect fear of the context rather than fear to the tone. This exclusion criterion has been used previously in this laboratory (e.g. Kim, McNally, & Richardson, 2006). Four rats were affected by this criterion. Thus final group sizes were: Group Control (n = 13); group
Over – 2 (n = 12); group Over – 4 (n = 12); group Over – 8 (n = 11). Data were analysed by means of planned orthogonal contrasts. The Type 1 error rate (α) was controlled at 0.05 for each contrast tested (Hays, 1972).
Results and Discussion
Stage I. Figure 6 shows the mean and standard error of the mean (SEM) levels of freezing during the three stages of the experiment. At the end of Stage I training there was significantly more freezing during CS presentations than during the pre-CS period
(F (1, 47) = 91.2, p < 0.05) (mean pre-CS = 11.2%, SEM = 2.1; mean CS = 48.9%,
SEM = 3.1).
Stage II. Freezing to the compound CS during Stage II is shown in the central panel of Figure 6. There was significantly more freezing during CS presentations (mean =
46.4%, SEM = 3.1) than during the pre-CS period (mean = 5.3%, SEM = 1.8) (F (1, 34)
= 130.4, p < 0.05). There was evidence for summation in freezing during the first Stage
II trial: freezing was significantly higher to the first presentation of the compound CS as compared to the average responding observed to the individual CSs at the end of Stage I
(F (1, 34) = 16.4, p < 0.05). It is worth noting that this is a liberal criterion for assessing summation (e.g., Aydin & Pearce, 1997; Lattal & Nakajima, 1998; Rescorla, 1997).
However a similar significant difference in freezing was observed when the first compound trial was compared to the final Stage I presentation of the tone (F (1, 34) =
12.9, p < 0.05) or flashing light CS (F (1, 34) = 12.2, p < 0.05). Nevertheless, it is
106 Chapter 4. General Discussion
Table 1. Design of Experiment 1
Group Stage I Stage II Test
Control Flash +, Tone + ______Tone
Over - 2 Flash +, Tone + Flash/Tone+ (2) Tone
Over - 4 Flash +, Tone + Flash/Tone+ (4) Tone
Over - 8 Flash +, Tone + Flash/Tone+ (8) Tone
Stage I Stage II Test
100 100 Over 2 75 Over 4 75 Over 8 50 50
25 25 PERCENT FREEZING PERCENT 0 0 Tone Flash 1 2 3 4 5 6 7 8 CTRL OVER OVER OVER 248 CS Trial GROUP
Figure 6. Mean CS freezing during each Stage of Experiment 1. The left panel shows mean freezing to the tone and flashing light on Days 5 and 6, respectively, of Stage I conditioning. The centre panel shows mean freezing to compound presentations for each of the three groups that received overexpectation training. The right panel shows mean freezing averaged over the 4 test presentations of the tone.
107 Chapter 4. General Discussion
possible that this increased level of freezing reflected additional fear learning resulting from conditioning trials on Day 5 and/or Day 6 rather than summation.
Inspection of the middle panel of Figure 6 indicates that freezing decreased as a function of the number of Stage II trials. There were no significant differences between groups in levels of freezing across the first two presentations of the compound (Fs (1,
32) < 1, ps > 0.05). There was a significant decrement in freezing across these first two trials (F (1, 32) = 18.5, p < 0.05) which did not interact with differences between groups
(Fs (1, 32) < 3.7, p > 0.05). The linear decrease in freezing across the first four Stage II trials for groups Over – 4 and Over – 8 was also significant (F (1, 21) = 17.7, p < 0.05).
These two groups did not differ over these four trials, and the between group contrast did not interact significantly with the linear decline in freezing (Fs (1, 21) < 1, ps >
0.05). Furthermore, the decrease in freezing across the eight Stage II trials for group
Over – 8 was also significant (F (1, 10) = 26.0, p < 0.05). The decrement in freezing observed during Stage II training could be evidence for overexpectation within-session but it may also be due to disruptive influences of the footshock US on expression of freezing.
Test. The data of primary interest are those from test and are shown in the right-hand panel of Figure 6. Inspection of the panel confirms the presence of overexpectation which, surprisingly, was inversely related to the amount of Stage II training. Levels of pre-CS freezing on test were low (group Control mean = 13.3%, SEM = 4.5; group
Over – 2 mean = 10.2%, SEM = 2.1; group Over – 4 mean = 15.6%, SEM = 4.5; group
Over-8 mean = 18.2%, SEM = 4.8) and did not differ between groups (Fs (1, 44) < 1.9, ps > 0.05). There was significantly more freezing during CS presentations compared to the pre-CS period, averaged across groups (F (1, 44) = 103.6, p < 0.05). The analysis showed that among the experimental groups receiving Stage II training, freezing was
108 Chapter 4. General Discussion
significantly and inversely related to the amount of Stage II training (F (1, 44) = 5.7, p <
0.05), implying that overexpectation was most pronounced with fewer Stage II trials.
This was confirmed by the findings that group Over - 2 (F (1, 44) = 4.3, p < 0.05), but not groups Over – 4 (F (1, 44) = 1.6, p > 0.05) and Over – 8 (F (1, 44) < 1, p > 0.05), differed significantly from group Control.
These results confirm the presence of overexpectation of conditioned fear (Blaisdell et al., 2001; Kremer, 1978; McNally et al., 2004a; Rescorla, 1970). They show, that under present conditions, overexpectation is maximal after two compound - shock pairings and is reduced with further Stage II training. The loss of overexpectation with over-training is unexpected from the perspective of error-correcting learning rules, which predict either no effect of number of trials (if the Stage II negative prediction error was fully corrected by learning from the first two compound trials) or an effect in the opposite direction (if further error-correction was possible beyond the first two compound trials). This result also contrasts with the results of Kremer (1978) who, using a conditioned suppression design, reported that 16, as opposed to 1, 4, or 8, Stage
II trials produced optimal overexpectation. Possible theoretical explanations for the over-training effect will be discussed in the following chapter. However, for present purposes, the most important result of this experiment was the finding that overexpectation could be detected after 2 compound - shock pairings.
It should be noted that the choice of control group could have opened this experiment to the criticism that any reduction in freezing detected in the overexpectation groups relative to the control group may have been due to US habituation rather than overexpectation. Such an argument would posit that during
Stage II training, rats habituated to the shock, reducing its aversive value and causing a devaluation effect (see, for example, Rescorla, 1973). In this case, it would simply be
109 Chapter 4. General Discussion
the shock presentations themselves, rather than their pairing with the compound stimulus, that would produce a decrement in freezing on test. In this case, a control group exposed to continued tone – shock or flash – shock pairings, or unsignalled shock presentations during Stage II may be expected to show a similar reduction in freezing to that shown by the overexpectation groups.
It is worth noting that Rescorla (1970), Kremer (1978), and Blaisdell et al. (2001) all used control groups that continued to receive shock presentations during Stage II conditioning, and demonstrated overexpectation relative to these groups. Rescorla
(1970) and Kremer (1978) found that, on test, these control groups did not differ significantly in conditioned suppression to a control group analogous to that used in the present experiment. Kehoe & White (2004) also found overexpectation of rabbit eyeblink conditioning using a control group that received continued training of elemental stimuli during Stage II. Thus, continued shock presentation during Stage II appears neither to abolish the overexpectation effect nor to alter responding relative to a control group that receives no Stage II treatment. These findings are inconsistent with the occurrence of US habituation and thus suggest that a control group that receives no
Stage II treatment is sufficient for detecting and demonstrating overexpectation. Within- subjects demonstrations of overexpectation in appetitive conditioning (Lattal &
Nakajima, 1998; Rescorla; 1999; 2006; 2007) also argue against a role for US habituation, because such habituation would equally affect performance to the control and target stimuli. Nevertheless, it could be argued that given the unique parameters and measure used by the present study, this possibility should still be considered and controlled for. However, if rats did habituate to the shock during Stage II training, this process should have continued and strengthened with additional shock presentations.
Thus, the decrement in freezing on test should have been equal or greater in group Over
110 Chapter 4. General Discussion
– 8 than in group Over – 2. Given that group Over – 8 actually froze more than group
Over – 2, the hypothesis that the effects observed were due to shock habituation resulting in US devaluation can be rejected, along with the need for further control groups.
Experiment 2
The aim of Experiment 2 was to test whether the benzodiazepine partial inverse agonist FG7142 modulates the expression of fear overexpectation as it does the expression of fear extinction. FG7142 reduces the ability of GABA to induce chloride flux at any GABAA receptor containing the benzodiazepine binding site. Recall that benzodiazepine binding sites are expressed on GABAA receptors containing α1, α2, α3, or α5 subunits (Mohler et al., 2002). FG7142 has greatest affinity for receptors containing α1 subunits. Its affinity at these receptors is approximately 3.6 times higher than at receptors containing α2 subunits, 5.4 times higher than at receptors containing α3 subunits, and 24 times higher than at receptors containing α5 subunits. The EC50 of
FG7142 is, accordingly, lowest at receptors containing α1 subunits. Its EC50 at α2, α3, and α5 subunit-containing receptors is approximately 3.70, 7.45, and 10.5 times higher, respectively. At concentrations of GABA capable of eliciting a 20% maximal chloride current, sufficiently high doses of FG7142 can decrease chloride currents by 43%, 34%,
37%, and 30% in α1, α2, α3, and α5 subunit-containing receptors respectively (Atack et al., 2005).
While FG7142 is a short-acting drug, its actions appear to be sufficiently sustained for it to be appropriate for use 15 min prior to the 18 min test session used in the present experiments. In mice, FG7142 binding is maximal 30 min after an i.p. injection of 40 mg/kg and returns to baseline between 120 and 240 min post-injection (Dawson et al., 111 Chapter 4. General Discussion
2006). Anxiogenic effects that have been reported in rats and humans appear to last approximately 30 min (Evans & Lowry, 2007). Another important neurochemical effect, increases in mPFC dopamine, which may be relevant to FG7142 modulation of extinction and overexpectation shows a similar time course. Bassareo, Tanda,
Petromilli, Giua, and Di Chiara (1996) found that 10, but not 5, mg/kg FG7142 caused an increase in dopamine levels in the mPFC of rats that was significant, relative to baseline levels, 20 and 40 min after i.p. injection. However, dopamine levels had returned to baseline by 1 hr after injection.
Experiment 2 studied whether expression of fear overexpectation could be disrupted by administering 10 mg/kg FG7142 15 min prior to test. At this dose, systemic FG7142 recovers freezing in rats after fear extinction (Harris & Westbrook, 1998) and infantile amnesia (Kim et al., 2006; Kim & Richardson, 2007), enhances avoidance of the open arms of a plus maze in rats (Atack et al., 2005; Cole, Hillmann, Seidelmann, Klewer, &
Jones 1995; Pellow & File, 1986) and mice (Rodgers, Cole, Aboualfa, & Stephenson,
1995), eliminates renewal and spontaneous recovery of magazine entries after appetitive extinction (Delamater, Campese, & Westbrook, In Press) and causes increased dopamine turnover in the rat mPFC (Tam & Roth, 1985; Bassareo et al., 1996; Murphy,
Arnsten, Jentsch, & Roth 1996). The design is shown in Table 2. All rats received Stage
I conditioning as in Experiment 1. Two groups received compound conditioning identical to that used in group Over - 2 of Experiment 1. One of these groups was injected with FG7142 15 min before test. The other was injected with vehicle. These groups were compared with a control group trained similarly to that used in Experiment
1 and injected with vehicle before test.
112 Chapter 4. General Discussion
Method
Subjects and Apparatus
Subjects were 33 experimentally naive male Wistar rats obtained, housed, and handled as in Experiment 1. The apparatus was the same as that described for
Experiment 1.
Drugs
FG7142 (N-methyl-β-carboline-3-carboxymide; Sigma-Aldrich Chemical Company,
Sydney, Australia) was suspended at a concentration of 10 mg/ml in saline (0.9% w/vol) using 1 drop of Tween 80 per 5 ml saline. This suspension, or the vehicle (saline plus
Tween 80) was administered in a volume of 1 ml/kg by s.c. injection into the dorsal region of the neck.
Procedure
Pre-exposure. In Experiment 1, rats were exposed to the conditioning chambers after experiencing several signalled shocks in that context. These exposures were intended to extinguish context fear, so that responses to the tone CS on test could be measured against a background of low context fear. If FG7142 recovers extinguished context fear, this could confound detection of its effects on overexpectation during the test session.
Therefore, in this experiment, we sought to reduce context conditioning in the first place by pre-exposing the rats to the context before the beginning of Stage I conditioning, using two 20 min sessions (one morning session and one afternoon session). These context pre-exposure sessions were repeated again three days later. Stage I conditioning was commenced the following day.
Stage I. Procedures were identical to those described for Experiment 1 except that afternoon context exposures were carried out on days 3 (10 min), 5 (20 min), and 6 (10 min), rather than on Days 4, 5, and 6 as in Experiment 1. Two rats failed to show >30%
113 Chapter 4. General Discussion
freezing to the tone on at least one of either Days 3 and 5, and were excluded from further procedures.
Stage II. Rats assigned to the control group were transported to the laboratory and received approximately 30 s of handling. Remaining rats underwent a procedure identical to that described for group Over – 2 in Experiment 1.
Test. Rats that underwent Stage II conditioning were administered either vehicle or
FG7142 15 min prior to the beginning of the test session. Rats in the control group received vehicle. Testing procedures were identical to those described for Experiment 1.
Data from one rat from group Over – FG7142 was excluded due to it freezing more than
50% of the time during the first 3 min of context exposure. Thus, final group sizes were:
Group Over – Vehicle (n = 11); group Over – FG7142 (n = 10); group Control (n = 9).
Results and Discussion
Stage I. The mean and SEM levels of freezing across the three stages of the experiment are shown in Figure 7. During the final CS presentations at the end of Stage
I training there was significantly more freezing during CS presentations as compared to the pre-CS period (mean pre-CS freezing = 11.2%; SEM = 3.1) (F (1, 29) = 81.4, p <
0.05).
Stage II. Due to a recording failure, context freezing data from three rats was unavailable. However, the freezing of the remaining 18 rats during the first three minutes of context exposure was 5.5% (SEM = 2.4%). This was significantly lower than the mean freezing to the CS presentations shown by these 18 rats (F(1, 17) = 75.8, p <
0.05). In contrast to Experiment 1, there was no evidence for summation in freezing during the first Stage II trial: freezing during the first presentation of the compound CS was not significantly different to the average responding observed to the individual CSs at the end of Stage I (F (1, 20) = 2.1, p > 0.05). As in Experiment 1, there was a
114 Chapter 4. General Discussion
Table 2. Design of Experiment 2
Group Stage I Stage II Test
Control Flash +, Tone + ______Veh - Tone
Over - Veh Flash +, Tone + Flash/Tone+ Veh - Tone
Over – FG7142 Flash +, Tone + Flash/Tone+ FG7142 - Tone
Stage I Stage II Test 100 100
75 75
50 50
25 25
PERCENT FREEZING PERCENT 0 0 Tone Flash 1 2 CTRL OVER OVER VEH FG7142 CS Trial
Figure 7. Mean CS freezing during each Stage of Experiment 2. Left panel shows CS freezing during days 5 and 6 of Stage I conditioning. Test data shows mean freezing averaged over all 4 tone presentations.
115 Chapter 4. General Discussion
significant decrease in freezing between the first and second CS presentation (F(1, 20) =
10.2, p < 0.05).
Test. Test data are shown in the left panel of Figure 7. Inspection of the panel suggests the presence of overexpectation among group Over – Vehicle and an attenuation of this overexpectation among group Over – FG7142. The analysis confirmed this. On test there was significantly more freezing during CS presentations as compared to the pre-CS period (F (1, 27) = 200.5, p < 0.05) (mean pre-CS freezing =
8.3%; SEM = 4.5). There were no significant differences between groups in levels of pre-CS freezing (group Control mean = 12.8%, SEM = 5.0; group Over – Vehicle mean
= 5.6%, SEM = 1.8; group Over – FG7142 mean = 7.2%, SEM = 4.3) (Fs (1, 27) < 1.6, ps > 0.05). There was a significant overexpectation effect because there was a significant difference in freezing between group Control and group Over – Vehicle (F
(1, 27) = 11.6, p < 0.05). Importantly, there was also a significant difference between group Over – Vehicle and group Over – FG7142 (F (1, 27) = 4.8, p < 0.05), indicating that pre-treatment with FG7142 prior to test attenuated expression of overexpectation.
It is possible that the difference between group Over – Vehicle and group Over –
FG7142 was due not to FG7142 attenuating expression of overexpectation, but rather to it attenuating expression of any extinction learning which might have occurred across the four non- reinforced CS presentations on test (Harris & Westbrook, 1998). To further examine this possibility, test performances across CS presentations were analysed (Figure 8). There was no significant linear decrease in freezing across test presentations of the CS (F (1, 27) < 1, p > 0.05). There was also no significant interaction between the between-groups contrasts analysed above and change in freezing across CS presentations (Fs (1, 27) < 1, ps > 0.05).
116 Chapter 4. General Discussion
100
75 Control - Vehicle Over - FG7142
50 Over - Vehicle
25
PERCENT FREEZING PERCENT 0 1 2 3 4 Trial
Figure 8. Mean freezing to each tone presentation during the test session in Experiment
2.
117 Chapter 4. General Discussion
The absence of any significant extinction across CS presentations, as well as of any interactions of the effects of FG7142 with CS presentations, argue strongly against the possibility that the effects of FG7142 on test were due to effects on extinction. Rather, the results from this experiment show that conditioned freezing which has been reduced by overexpectation can be recovered by injection of FG7142. The recovery of freezing after overexpectation seen in rats injected with FG7142 before test adds to the evidence from experiments using lick suppression (Blaisdell et al., 2001) and appetitively- motivated magazine entries (Rescorla, 2006; 2007) that the decrement in conditioned responding caused by overexpectation training is apparently neither permanent nor irreversible. This is consistent with the general proposal that the expression of overexpectation requires the partial masking of the CS - US association learned during
Stage I conditioning. More specifically, it suggests that in fear conditioning, this mask is mediated by GABAA receptor activation. As treatment with FG7142 before testing also promotes recovery of freezing after extinction of fear to an auditory CS (Harris &
Westbrook, 1998), this finding is consistent with common mechanisms contributing to both overexpectation and extinction.
Experiment 3
In previous demonstrations of recovery of freezing by FG7142 (Harris & Westbrook,
1998; Kim et al., 2006, Kim & Richardson, 2007), a 10 mg/kg dose was consistently effective. Harris and Westbrook (1998) also found significant effects on recovery from fear extinction at a 5, but not 2.5 mg/kg. Kim et al. (2006) found significant effects on infantile amnesia in 26 day old rats at 5, but not at 1 mg/kg, while Kim and Richardson
(2007) found that neither the 1 nor 5 mg/kg doses were effective in recovering freezing from infantile amnesia. The purpose of Experiment 3 was to assess the dose-response
118 Chapter 4. General Discussion
properties of FG7142 on expresssion of overexpectation. The doses of FG7142 used, 0,
1, or 10 mg/kg, were chosen based on this past research showing that 10 mg/kg was consistently effective whereas 1 mg/kg was consistently ineffective, in modulating expression of freezing. If freezing after overexpectation shows similar differential sensitivity to modulation by the low vs. high dose of FG7142, then rats administered the
1 mg/kg dose should show similar levels of freezing as rats injected with vehicle, while rats injected with 10 mg/kg FG7142 should again show recovery of freezing. The design of the experiment is shown in Table 3.
Method
Subjects, Apparatus, Drugs
Subjects were 63 experimentally naive male Wistar rats obtained, housed, and handled as in previous experiments. The apparatus and was identical to that used previously. FG7142 was suspended at a concentration of either 1 mg/ml or 10 mg/ml in saline (0.9% w/vol) using 1 drop of Tween 80 per 5 ml saline. This suspension, or the vehicle was administered s.c. in a volume of 1 ml/kg.
Procedure
Stage I. Rats received pairings of tone and flashing light with shock in Stage I as described in previous experiments. In this experiment Stage I training was reduced from six to four days so that at the end of training rats had received 2 pairings of each CS with shock. Rats received 20 min exposures to the conditioning chambers on the afternoons of Days 3 and 4. Fifteen rats failed to show freezing levels over 30% to the tone on Day 3 and were excluded from further procedures.
Stage II. Compound conditioning procedures were identical to those used in
Experiment 2.
119 Chapter 4. General Discussion
Test. Rats that had received Stage II training were injected with either 1 mg/kg
FG7142, 10 mg/kg FG7142 or vehicle. Control rats were injected with vehicle. Fifteen min after injection, rats were tested as described previously. Data from 2 rats from group Over – 0 mg/kg, 3 rats from group Over – 1 mg/kg, and 3 rats from group over –
10 mg/kg were excluded as they showed freezing levels of >50% during the first three minutes of context exposure. Final group sizes were: Group Control (n = 12); group
Over - 0 mg/kg (n = 11), group Over - 1 mg/kg (n = 8), and group Over – 10 mg/kg (n =
9).
Results and Discussion
Stage I. The mean and SEM levels of freezing across the three stages of the experiment are shown in Figure 9. During the final CS presentations at the end of Stage
I training there was significantly more freezing during CS presentations as compared to the pre-CS period (mean pre-CS freezing = 11.8%; SEM = 2.7) (F (1, 39) = 106.7, p <
0.05).
Stage II. During Stage II training there was significantly more freezing during CS presentations as compared to the pre-CS period (mean pre-CS freezing = 5.8%, SEM =
1.9) (F (1, 27) = 97.5, p < 0.05). There was no evidence for summation in freezing during the first Stage II trial: freezing during the first presentation of the compound CS was not significantly different to the average responding observed to the individual CSs at the end of Stage I (F (1, 27) = 2.3, p > 0.05). There was a significant difference in freezing across the two Stage II CS presentations, so that again freezing decreased significantly between the first and second CS presentation (F (1, 27) = 44.8, p < 0.05).
Test. Inspection of the test data suggests the presence of overexpectation among group Over – Vehicle and an attenuation of overexpectation by FG7142 in a dose- dependent manner. The analysis confirmed this observation. On test there was
120 Chapter 4. General Discussion
Table 3. Design of Experiment 3
Group Stage I Stage II Test
Control Flash +, Tone + ______Veh - Tone
Over – 0 mg/kg Flash +, Tone + Flash/Tone+ Veh - Tone
Over – 1 mg/kg Flash +, Tone + Flash/Tone+ 1 mg/kg FG7142 - Tone
Over – 10 mg/kg Flash +, Tone + Flash/Tone+ 10 mg/kg FG7142 - Tone
Stage I Stage II Test
100 100
75 75
50 50
25 25 PERCENT FREEZING 0 0 Tone Flash 1 2 CTRL OVER OVER OVER 0110 CS trial FG7142 mg/kg
Figure 9. Mean CS freezing during each Stage of Experiment 3. Left panel shows CS freezing during days 3 and 4 of Stage I conditioning. Test data shows mean freezing averaged over all 4 tone CS presentations.
121 Chapter 4. General Discussion
significantly more freezing during CS presentations as compared to the pre-CS period
(F (1, 36) = 138.7, p < 0.05) (mean pre-CS freezing = 11.8%; SEM = 1.8). Freezing during the first 3 min of context exposure was significantly lower in group Control
(mean = 5.3%, SEM = 2.5) than in the overexpectation groups (Over – 0 mg/kg mean =
13.5%, SEM = 2.9; Over – 1 mg/kg mean = 18.2%, SEM = 5.8; Over – 10 mg/kg mean
= 12.8%, SEM = 3.0) (F (1, 36) = 6.4, p < 0.05). The overexpectation groups did not differ from each other in levels of pre-CS freezing (Fs (1, 36) < 1.1, ps > 0.05).
There was evidence for overexpectation because overall, averaged across doses of
FG7142, groups receiving Stage II training showed significantly less freezing to the CS on test than did group Control (F (1, 36) = 4.3, p < 0.05). There was also evidence that
FG7142 attenuated expression of this overexpectation in a dose-dependent manner.
Group Over – 10 mg/kg showed significantly greater freezing than groups Over – 1 mg/kg and Over – 0 mg/kg (F (1, 36) = 6.9, p < 0.05), which did not differ from each other (F (1, 36) < 1, p > 0.05). A simple comparison of group Over – 10 mg/kg and group Over – 0 mg/kg, external to the family of planned orthogonal contrasts, likewise revealed a significant attenuation of overexpectation (t (18) = -2.4, p < 0.05).
To further confirm that any effects of FG7142 on expression of overexpectation were not secondary to effects on any extinction that might have occurred during test, performance across the four test CS presentations were analysed (data not shown).
There was no overall significant linear decrease in freezing across CS presentations (F
(1, 36) < 1, p > 0.05) and there were no interactions between the contrasts assessing differences between groups with the contrast assessing linear decreases in freezing on test (Fs (1, 36) < 2.2, ps > 0.05). The results of this experiment replicate the finding, reported in Experiment 2, that FG7142 attenuates expression of overexpectation and also show that this attenuation is dose-dependent. It also suggests that the sensitivity of
122 Chapter 4. General Discussion
overexpectation to blockade by FG7142 is similar to that of other decremental processes that have been investigated: extinction and infantile amnesia.
Experiment 4
While Experiments 2 and 3 have shown that expression of overexpectation is blocked by 10 mg/kg FG7142, their designs do not permit selection between several possible mechanisms by which the drug has this effect. One possibility is that overexpectation leads to the imposition of a decremental process associated with the
CS, and expressed through GABAA receptors. However, another possibility is that the effect of FG7142 may be due to state-dependent memory. According to this possibility, injection of FG7142 prior to test produced a shift in internal state/context which prevented overexpectation learning from Stage II (drug-free) generalising to test (drug- state). For fear extinction there is some evidence for such generalisation decrement based on injections of chlordiazepoxide prior to extinction training. Bouton et al. (1990) showed that the amnesia for extinction that followed extinction training under the influence of the benzodiazepine could be relieved by administration of the drug prior to test. As rats were conditioned drug-free and subjected to extinction training in the presence of the drug, manipulating the drug state on test was analogous to a test for
ABA vs. ABB renewal. In Experiments 2 and 3, Stage II training was conducted drug- free, and test was conducted in the presence of either FG7142 or vehicle. It could be argued that this manipulation of drug state was analogous to an AAB vs AAA renewal design. It is worth noting that there is little evidence for such effects of FG7142 in the literature. For example, Harris and Westbrook (1998) showed that the effects of
FG7142 on fear extinction cannot be attributed to such state-dependent memory processes.
123 Chapter 4. General Discussion
A third possibility is that Stage II training produced partial erasure of the fear memory and that FG7142 simply enhanced fear and/or performance of conditioned freezing. If this is the case, FG7142 may have augmented freezing regardless of the presence versus absence of Stage II training. The designs of Experiments 2 and 3 did not permit assessment of this possibility. Indeed, reports of putative anxiogenic effects of FG7142 lend intuitive appeal to it. FG7142 has been shown to: 1) enhance effects of low-level punishment on suppression of drinking in rats (1.8 mg/kg) (Corda, Blaker,
Mendelson, Guidotti, & Costa, 1983); and, 2) enhance suppression of food-reinforced bar-pressing upon offset of a stimulus that signals absence of random punishment of bar presses by shock in rats (4 and 8 mg/kg) (Thiebot, Dangoumau, Richard, & Puech,
1991). Together these two reports suggest that FG7142 enhances avoidance behaviour, consistent with it being anxiogenic. Furthermore, FG7142 reduces time spent in open arms of an elevated plus maze in rats at doses of 10-100 mg/kg (Atack et al., 2005; Cole et al., 1995; Pellow & File, 1986) and reduces numbers of open arm entries in mice at
10 mg/kg (Rodgers et al., 1995). Finally, humans administered doses of 200-400 mg
FG7142 have reported symptoms similar to a panic attack (Dorow, Horowski,
Paschelke, & Amin, 1983).
Despite these reports of its anxiogenic effects, there is no evidence that FG7142 potentiates Pavlovian fear responses that have not been subject to some decremental process. Indeed, there is some evidence that, in juvenile rats, FG7142 can aid the recall of inhibition of fear when inhibitory training precedes fear conditioning. Kim et al.
(2006) found that 10 mg/kg FG7142 enhanced expression of latent inhibition in 26 day old rats. Rats were exposed to 45 presentations of a noise CS alone, then exposed to CS- shock pairings the following day, and tested a day after conditioning. These rats showed less freezing to the CS than non-pre-exposed controls. If injected with FG7142 before
124 Chapter 4. General Discussion
test, this latent inhibition was enhanced. These results suggest that FG7142 does not potentiate fear CRs. Instead, they suggest that it has a general effect of facilitating retrieval of memories degraded by the passage of time, or subject to interference by more recent memories, regardless of whether or not these older memories inhibit or excite fear. Nevertheless, the overexpectation effect demonstrated in Experiments 1 - 3 was demonstrated in adult rats. The effects of FG7142 in adult rats may be different to its effects in the juvenile rats used by Kim et al. (2006). If FG7142 generally increased freezing, it should do regardless of Stage II training. To test this possibility, two control groups were employed in Experiment 4, one injected with vehicle on test and one injected with FG7142. The design of Experiment 4 is shown in Table 4.
Method
Subjects, Apparatus, Drugs
Subjects were 56 experimentally naive male Wistar rats obtained, housed, and handled as in previous experiments. The apparatus and was identical to that used previously. Drugs were as described for Experiment 2.
Procedure
Stage I. The Stage I conditioning procedure was identical to that described for
Experiment 3. Fifteen rats failed to freeze >30% during the tone presentation on Day 3 and were excluded from further procedures.
Stage II. Compound conditioning procedures were identical to those used in
Experiments 2 and 3. 15 min before compound conditioning, rats received injections of either 10 mg/kg FG7142 or vehicle. Rats in the control group were also injected with either 10 mg/kg FG7142 or vehicle and briefly handled 15 min later.
Test. Fifteen min before test, all rats received the same injection that they had received prior to Stage II conditioning (or handling in the control groups). Test
125 Chapter 4. General Discussion
procedures were identical to those described for previous experiments. Data from one rat in group Over - Veh was excluded because it froze >50% during the first 3 min context exposure. Final group sizes were: Group Control – Veh (n = 10); group Control
– FG (n = 10); group Over – Veh (n = 10); group Over – FG (n = 10).,
Results and Discussion
Stage I. The mean and SEM levels of freezing across the three stages of the experiment are shown in Figure 10. During the final CS presentations at the end of
Stage I training there was significantly more freezing during CS presentations as compared to the pre-CS period (mean pre-CS freezing = 12.5%; SEM = 1.6) (F (1, 39)
= 371.3, p < 0.05).
Stage II. During Stage II training there was significantly more freezing during CS presentations as compared to the pre-CS period (mean pre-CS freezing = 9.8%, SEM =
3.1) (F (1, 18) = 93.2, p < 0.05) and there were no differences between groups in pre-CS levels of freezing (F (1, 18) < 1, p > 0.05). As in Experiments 2 and 3, there was no significant evidence for summation on the first presentation of the compound CS (F (1,
18) = 3.2, p > 0.05), and this contrast did not interact with drug treatment (F (1, 18) < 1, p > 0.05). There was a significant difference in freezing across Stage II, so that again freezing decreased significantly between the first and second CS presentation (F (1, 18)
= 12.4, p < 0.05). However, FG7142 had no effect on freezing to the compound CS, and did not interact with the within-subjects decline in freezing between the first and second compound CS presentation (Fs (1, 18) < 1, ps > 0.05). This is consistent with the hypothesis that FG7142 has no general effect on conditioned freezing.
Test. Mean freezing to the tone during the test session is shown in the right panel of
Figure 10. Freezing during the first 3 min of the test session was 3.6% (SEM = 1.9) in group Control -Veh, 9.0% (SEM = 4.5) in group Control – FG7142, 10.1% (SEM = 3.6)
126 Chapter 4. General Discussion
Table 4. Design of Experiment 4
Group Stage I Stage II Test
Control - Veh Flash +, Tone + Veh - handling Veh - Tone
Control - FG Flash +, Tone + FG7142 - handling FG7142 - Tone
Over – Veh Flash +, Tone + Veh - Flash/Tone+ Veh - Tone
Over – FG Flash +, Tone + FG7142 - Flash/Tone+ FG7142 - Tone
Stage I Stage II Test 100 Vehicle Vehicle 100 FG7142 FG7142 75 75
50 50
25 25 PERCENT FREEZING PERCENT 0 0 Tone Flash 1 2 Control Over CS Trial GROUP
Figure 10. Mean CS freezing during each stage of Experiment 4. Left panel shows freezing on days 3 and 4 of Stage I. Test data is shown averaged over all 4 tone presentations.
127 Chapter 4. General Discussion
in group Over - Veh, and 14.3% (SEM = 4.9) in group Over – FG7142. Context freezing was not significantly affected by Stage II conditioning or FG7142 (Fs (1, 36) <
2.3, ps > 0.05). Freezing was significantly greater during tone presentations than during the first 3 min of context exposure (F (1, 36) = 251.2, p < 0.05).
Inspection of the test data suggests the presence of overexpectation in group Over –
Vehicle but not group Over – FG7142. Group Over – Vehicle differed significantly from groups Control – Vehicle and Control – FG7142 (F (1, 36) = 13.5, p < 0.05).
There was however no such significant difference between group Over – FG7142 and groups Control – Vehicle and Control – FG7142 (F (1, 36) = 1.9, p > 0.05). There was also no significant difference between groups Control – Vehicle and Control – FG7142
(F (1, 36) < 1, p > 0.05). A simple comparison of group Over – FG7142 and group Over
– Vehicle (external to the family of planned orthogonal contrasts) likewise revealed a significant attenuation of overexpectation (t (18) = -1.9, p < 0.05 one-tailed). There was an overall significant linear decrease in freezing across test CS presentations (F (1, 36)
= 10.2, p < 0.05) but there were no interactions between the contrasts assessing differences between groups with the contrast assessing decreases in freezing on test (Fs
(1, 36) < 1, ps > 0.05).
The results of this experiment confirm the presence of overexpectation of fear and extend the findings from Experiments 2 and 3 that FG7142 attenuates expression of overexpectation. It is clear from these results that FG7142 did not affect freezing in the control group. This suggests that the effect of FG7142 on expression of overexpectation is not due to a general augmentation of fear or freezing. The evidence against a state- dependent memory effect of FG7142 is somewhat weaker. The evidence for an attenuation of expression of overexpectation by FG7142 is based on a liberal criterion of significance (a direct comparison external to the family of planned orthogonal
128 Chapter 4. General Discussion
contrasts, which was only significant at the one-tailed level). Nevertheless, if the effect of FG7142 in Experiments 2 and 3 were due to state-dependent memory, this difference should not have been observed. Furthermore, if state-dependent memory was the mechanism, then a significant overexpectation effect would have been predicted in group Over – FG7142. This was not the case. Group Over – FG7142, unlike group Over
– Vehicle, did not differ significantly from the two control groups. Thus, these results do not support the possibility that the effect of FG7142 on overexpectation is due to state-dependent learning or memory. However, these results do suggest that the effect of
FG7142, under the present conditions, is small.
To further elucidate the effect of FG7142 on the expression of overexpectation, all data from rats in groups Control (Group Control [Experiments 2 and 3]; Group Control
– Vehicle [Experiment 4]), groups Over – Vehicle (group Over – Vehicle [Experiment 2 and 4]; group Over – 0 mg/kg [Experiment 3]), and group Over – 10 mg/kg FG7142
(group Over – FG7142 [Experiments 2 and 4]; group Over – 10 mg/kg [Experiment 3]) were combined and 95% confidence intervals for their corresponding population means estimated. There were procedural differences between experiments included in this analysis, shown in Figure 11, but the important point is that rats were tested under identical conditions and this enables estimation of the relevant effect sizes. Consistent with suggestions of Cummings and Finch (2005), data are expressed in raw score units
(percentage freezing on test across all CS presentations). Inspection of Figure 11 confirms that overexpectation training reduces freezing on test and that FG7142 increases freezing after overexpectation training.
Analysis of the pooled data showed, with 95% confidence, that the population mean for groups Control was contained within the interval 56.2% - 68.2%. Likewise, with
95% confidence, the population mean for groups Over - Vehicle was contained within
129 Chapter 4. General Discussion
100
75
50
25
PERCENT FREEZING
0 Control Over Over Veh FG7142
Figure 11. Individual test data points with 95% confidence intervals for population means for groups Control, Over - Vehicle, and Over - FG7142 from subjects reported in
Experiments 2, 3, and 4.
130 Chapter 4. General Discussion
the interval 34.6% - 47.5%. Finally, with 95% confidence, the population mean for groups Over – FG7142 was contained within the interval 49.3% - 62.4%. This separation between the 95% confidence interval of group Over-Vehicle and group
Control demonstrates a significant difference (p < 0.05) between population means
(Cummings & Finch, 2005). Likewise, the seperation between the 95% confidence intervals for group Over – Vehicle and Over – FG7142 demonstrates a significant difference (p < 0.05) between population means (Cummings & Finch, 2005). Finally, the overlap in the 95% confidence intervals for group Over – FG7142 and group
Control suggests absence of difference between population means (Cummings & Finch,
2005). This supports the inference that overexpectation can be observed under the present conditions and that its expression is modulated by FG7142. However, while the
95% confidence intervals of groups Over – Vehicle and groups Over – FG7142 are separated, this separation is small. The direction of this difference is consistent with an alleviation of overexpectation by FG7142. However, the small size of this effect may render it difficult to detect. This provides a reason as to why the direct comparison between group Over – Vehicle and group Over – FG7142 was significant only at the one-tailed level in Experiment 4, but reached significance with more conservative tests in Experiments 2 and 3.
Experiment 5
The absence of an effect of FG7142 on freezing in the control group in Experiment 4 suggests that its effects on expression of overexpectation are not simply due to a generalised potentiation of freezing. However, this observation is an insufficient basis upon which to draw the conclusion that FG7142 has no generalised effects on freezing.
Firstly, the possibility that rats in the control group were freezing at ceiling levels was
131 Chapter 4. General Discussion
not controlled for in Experiment 4. If this were the case, potentiation of freezing could not be observed even if FG7142 were potentiating the underlying cause of the freezing
(fear). Secondly, it is possible that FG7142 has a generalised effect on freezing, but only when it is expressed at low or moderate levels, such as those shown by rats expressing overexpectation.
The present experiment addressed these possibilities. This design is summarised in
Table 5. Two groups were exposed to pairings of a tone with a low intensity shock (0.3 s, 0.3 mA). This was intended to condition levels of freezing similar to that seen to the tone after overexpectation training in Experiments 1 - 4. Of these groups, one was tested after an injection of 10 mg/kg FG7142 and the other after injection of vehicle. These rats were compared to rats conditioned with a high intensity shock (1 mA, 1 s). This group was included to confirm that rats in the low-shock/vehicle group were not freezing at ceiling levels.
Method
Subjects, Apparatus, and Drugs
Subjects were 32 experimentally naive male Wistar rats obtained, housed, and handled as in previous experiments. The apparatus and drugs were identical to those used in Experiments 2 and 4.
Procedure
Conditioning. On the morning of Day 1 rats were placed in the conditioning chambers and exposed to 2 tone-shock pairings, one beginning 3 min and 47 s after placement in the chamber, and the next beginning 8 min and 18 s after placement.
Group High - Vehicle were conditioned using a 1 mA, 1 s shock, while group Low –
Vehicle and Low – FG7142 were conditioned with a 0.3 mA, 0.3 s shock. Both shocks co-terminated with the tone. These shock parameters were chosen after extensive pilot
132 Chapter 4. General Discussion
studies which indicated that they produced reliable differences in levels of auditory fear conditioning of a similar magnitude as was observed using overexpectation training.
Rats were removed after a total of 14 min in the chambers. In the afternoon of the same day rats were placed in the chambers for a 10 min context exposure during which no CS or US events were scheduled.
Test. The following day, rats in groups Low were injected with either 10 mg/kg FG-
7142 or vehicle. Rats in the group High - Vehicle were injected with vehicle. Fifteen min after injection, rats were placed in the chambers and exposed to 6 presentations of the tone beginning 4 min after introduction into the chamber, with a 2 min and 31 s ISI.
Data from one rat from group Low – Vehicle and one from group High-Vehicle were excluded because these rats froze >50% during the first three minutes of context exposure. Final group sizes were: Group High – Vehicle (n = 10); group Low – Vehicle
(n = 10); group Low – FG7142 (n = 10).
Results and Discussion
The mean and SEM levels of freezing on test are shown in Figure 12. On test there was significantly more freezing during CS presentations as compared to the pre-CS period (F (1, 27) = 127.3, p < 0.05) (mean pre-CS freezing = 10.0%; SEM = 2.4). There was a significant difference between groups in pre-CS freezing, so that group High –
Vehicle (mean = 24.1%, SEM = 4.2) displayed significantly more pre-CS freezing than groups Low – Vehicle (mean = 4.1%, SEM = 2.3) and Low - FG7142 (mean = 1.8%,
SEM = 1.0) (F (1, 27) = 37.0, p < 0.05). There was no such difference between groups
Low – Vehicle and Low – FG7142 (F (1, 27) < 1, p > 0.05). Importantly, Group High –
Vehicle showed significantly more freezing during tone presentations than groups Low
– Vehicle and Low – FG7142 (F (1, 27) = 9.7, p < 0.05) whereas groups Low – Vehicle and Low – FG7142 did not differ significantly from each other (F (1, 27) < 1, p > 0.05).
133 Chapter 4. General Discussion
Table 5. Design of Experiment 5
Group Conditioning Test
High Shock Tone + (1 mA, 1 s) Veh - Tone
Low Shock - Veh Tone + (0.3 mA, 0.3 s) Veh - Tone
Low Shock - FG Tone + (0.3 mA, 0.3 s) FG7142 - Tone
100
75
50
25
PERCENT FREEZING 0 Low - Veh Low - FG High - Veh
Figure 12. Mean freezing to test CS presentations in Experiment 5. Data shown is averaged over all 6 tone presentations.
134 Chapter 4. General Discussion
These results show that injection of FG7142 does not act simply to increase the expression of fear or augment freezing. Injection of FG7142 prior to test did not increase freezing in rats conditioned with the lower magnitude footshock. The superior freezing on test shown by rats in group High - Vehicle shows that this failure to find any effect of FG7142 cannot be due to a ceiling effect on measurement. This, therefore, adds support to an interpretation of Experiments 2 - 4 which suggests that FG7142 did not alleviate overexpectation by generally increasing fear or freezing.
Experiments 6a and 6b
These experiments further characterised the effects of FG7142 on the expression of conditioned fear. The results obtained in Experiment 5 show that FG7142 does not simply increase fear or freezing on test. However, different levels of fear were produced in that experiment via manipulations of US magnitude whereas in an overexpectation design different levels of fear are produced via manipulations of CS predictiveness.
Specifically, an overexpectation design arranges that the summed associative strengths of the CSs exceed the amount of learning supported by the US during Stage II. In the present experiments different levels of fear were also produced via manipulations of the predictive status of the CS. These experiments studied the effects of FG7142 on the expression of fear which had been reduced by associative blocking. In a blocking design, subjects are first trained to fear CSA via pairings with shock. Then, a compound of CSA + CSB is arranged to signal the same shock. Tests reveal that less is learned about CSB compared to a control group not subject to Stage I training.
Overexpectation is a variant of the blocking design. The key difference is that in an overexpectation design, the target CS receives Stage I training whereas in a blocking design it does not. Therefore Stage II training in an overexpectation design involves
135 Chapter 4. General Discussion
negative predictive error whereas in a blocking design it does not. By contrast, the physical events experienced by the subjects during Stage II of a blocking and overexpectation design are otherwise similar. In both cases, a compound CS is paired with a footshock US and fear is later assessed. Studying the effects of FG7142 in a blocking design enables examination of the extent to which the effects of FG7142 on expression of fear to a CS are dependent on that CS being subject to a negative prediction error during Stage II. Experiment 6a used a two group, single factor design to confirm that blocking of fear learning could be observed under present conditions. The design is shown in Table 6. Experiment 6b used a four group design to study the effects of FG7142 on the expression of this blocked fear. The design is shown in Table 7.
Method
Subjects and Apparatus
Subjects were experimentally naive male Wistar rats (Experiment 6a N = 16;
Experiment 6b N = 32) obtained, housed, and handled as in previous experiments. The apparatus was as previously described.
Procedure
Experiment 6a. Pre-exposure. On Days 1 and 2 rats were placed in the conditioning apparatus and exposed to four 30 s presentations each of a flashing light (4Hz) and 82 dB, 750 Hz tone at an ISI of 30 s. The order of stimulus presentation was counterbalanced, and began 3 min after placement in the chambers. No shocks were delivered during these sessions. Rats were removed from the chambers and returned to their home cages 30 s after the end of the last CS presentation.
Stage I. On the mornings of Days 3 - 5 rats in the blocking group (n = 8) were placed in the conditioning chambers for 21 min and 30 s. During this time rats received four 30 s presentations of the flashing light co-terminating with 1 s, 0.5 mA footshock at an
136 Chapter 4. General Discussion
average inter-trial interval of 230 s. In the afternoons of these days, all rats in the blocking groups were exposed to the conditioning chambers for 10 min with no stimulus presentations. Rats in the control group (n = 8) were briefly handled once in the morning and once in the afternoon of each day of Stage I.
Stage II. On the morning of Day 6 all rats were placed in the conditioning chambers for 14 min. During this time, they received two 30 s compound presentations of the tone and flashing light co-terminating with shock. The first presentation occurred 5 min after placement in the chamber. The second presentation occurred 3 min later. In the afternoon, all rats received a 10 min exposure to the chambers.
Test. On Day 7 all rats were placed in the conditioning chambers and exposed to four 30 s presentations of the tone. The first CS presentation began 5 min after placement in the chambers, and further CS presentations followed at an ISI of 91 s.
Experiment 6b. Procedures were identical to those described for Experiment 6a, except that 15 min before the test procedure, rats in the blocking groups received s.c. injection of either 0, 1, or 10 mg/kg FG7142. Rats in the control group were injected with vehicle. Data from one rat from group Control and one rat from group Block – 0 mg/kg were excluded from analyses because these rats froze >50% during the first three minutes of context exposure during the test session. Final group sizes were: group
Control (n = 7), group Block – 0 mg/kg (n = 7), group Block – 1 mg/kg (n = 8), group
Block - 10 mg/kg (n = 8).
Results and Discussion
Experiment 6a
Stage I and II training proceeded uneventfully. The data of interest are those from test and the mean as well as SEM levels of freezing are shown in Figure 13. On test there was significantly more freezing during CS presentations as compared to the pre
137 Chapter 4. General Discussion
Table 6. Design of Experiment 6a
Group Stage I Stage II Test
Control __ Flash/Tone+ Tone
Blocking Flash + Flash/Tone+ Tone
100
75
50
25
PERCENT FREEZING 0 Control Block
Figure 13. Mean freezing to test CS presentations in Experiment 6a. Data shown is averaged over all 4 CS presentations.
138 Chapter 4. General Discussion
CS period (F (1, 14) = 26.4, p < 0.05) (mean pre-CS freezing = 12.3%; SEM = 2.8).
There was significantly more pre-CS freezing in group Block (mean = 20.8%, SEM =
3.8) than group Control (mean = 4.7%, SEM = 1.5) (F (1, 14) = 14.3, p < 0.05).
Importantly, there was evidence for the blocking of CS fear learning because Group
Control displayed significantly more freezing to the CS than group Block (F (1, 14) =
13.3, p < 0.05). This experiment therefore replicates the blocking effect with the same stimuli used in the overexpectation experiments and using similar compound conditioning and test procedures.
Experiment 6b
Stage I and II training proceeded uneventfully. The data of interest are those from test and the mean as well as SEM levels of freezing are shown in Figure 14. There were no differences between groups in pre-CS freezing (Fs (1, 26) < 1, ps > 0.05). On test there was significantly more freezing during CS presentations as compared to the pre-
CS period (F (1, 26) = 95.7, p < 0.05) (mean pre-CS freezing = 12.9%; SEM = 2.1).
Groups that received Stage I conditioning displayed less freezing to the CS than group
Control (F (1, 26) = 11.7, p < 0.05). There was no significant effect of FG7142 on the expression of blocking because group Block – 10 mg/kg did not differ from groups
Block – 0 mg/kg and Block – 1 mg/kg (F(1, 26) < 1, p > 0.05), which did not differ from each other (F(1, 26) = 1.1, p > 0.05). There was no significant within-subjects linear trend for freezing across the 4 CS presentations (F (1, 26) < 1, p > 0.05) and this contrast did not significantly interact with any of the between-groups contrasts mentioned above (Fs (1, 26) < 1.7, ps > 0.05).
These experiments confirm the associative blocking of fear conditioning. Experiment
6b also showed that expression of blocking of conditioned fear, unlike overexpectation and extinction, is unaffected by FG7142 at doses identical to those used in Experiment
139 Chapter 4. General Discussion
Table 7. Design of Experiment 6b
Group Stage I Stage II Test
Control __ Flash/Tone+ Veh - Tone
Block – 0 mg/kg Flash + Flash/Tone+ Veh - Tone
Block – 1 mg/kg Flash + Flash/Tone+ 1 mg/kg FG7142 - Tone
Block – 10 mg/kg Flash + Flash/Tone+ 10 mg/kg FG7142 -
Tone
100
75
50
25
FREEZING PERCENT 0 Control Block Block Block
0110 0
FG7142 (mg/kg)
Figure 14. Mean freezing to test CS presentations in Experiment 6b. Data shown is averaged over all 4 tone presentations.
140 Chapter 4. General Discussion
3. This failure of FG7142 to modulate responding to a blocked CS is especially informative. The procedures for Stage II training in the blocking and overexpectation designs were similar, so that for both cases a compound CS was paired with a footshock
US and fear was later assessed to the tone alone. The key difference was the nature of the Stage II prediction error. In the overexpectation design, Stage II predictive error was negative whereas in a blocking design this error was positive but negligible. Thus, these results are consistent with an account of extinction, overexpectation, and blocking that assumes that a negative prediction error results in an inhibitory memory that attenuates expression of the previously established excitatory conditioning memory, while small positive predictive error simply produces a weak excitatory memory. The inhibitory memory that results from negative prediction error is vulnerable to modulation of
GABAA receptors, and FG7142 therefore interferes with this expression. As blocking involves no inhibitory memory formation, and FG7142 has no general effect on fear expression, it has no effect on freezing to a blocked CS.
Within this theoretical framework, these results also replicate the finding of
Experiment 5 that FG7142 does not affect of expression of low to moderate levels of freezing. Freezing levels displayed by the blocking groups on test in Experiment 6b were very similar to those shown by overexpectation groups tested in the absence of
FG7142 in previous experiments. This adds weight to the contention that, in the overexpectation experiments, FG7142 did not simply potentiate the otherwise weak or moderate freezing exhibited to the tone after overexpectation.
Experiment 7
An additional way to examine the effect of FG7142 on expression of fear is to test its effect on freezing during a conditioning session. This approach allows investigation of
141 Chapter 4. General Discussion
several questions not answerable by the designs used in Experiments 5 and 6. Firstly, it allows examination of whether FG7142 affects fear expression when a CS is presented minutes, rather than a day, after its pairings with shock. This approach also allows investigation of another question which, to the author’s knowledge, has yet to be investigated: Does FG7142 affect the acquisition and/or consolidation of conditioned fear? This can be tested by observing freezing the following day in a drug free state.
This is of some relevance to Experiment 4, where rats in Group Over – FG received
FG7142 before compound – shock pairings during Stage II. If FG7142 alters the ability of shock to condition fear to a tone, this would have affected the learning resulting from
Stage II training, possibly in a way which would have confounded detection of state- dependent memory effects.
In Experiment 7, three groups of rats were exposed to 2 tone – shock pairings after injection of 0, 1 or 10 mg/kg FG7142. All rats were exposed to 6 non-reinforced tone presentations on each of the following two days to test long-term fear memory and its rate of extinction. The design of this experiment is summarised in Table 8.
Method
Subjects, Apparatus, and Drugs
Subjects were 23 male Wistar rats obtained, housed, and handled as in previous experiments. Apparatus and drugs were as described for Experiments 3 and 6b.
Procedure
Conditioning. On Day 1, rats were injected with 1 ml/kg of either 0 (group 0 mg/kg; n = 8), 1 (group 1 mg/kg; n = 7), or 10 (group 10 mg/kg, n = 8) mg/ml FG7142. 15 min later they were placed in the conditioning chambers for 14 min, during which they were exposed to 2 presentations of the tone co-terminating with a 1 s, 0.5 mA shock. The timing of tone presentations was as in Experiment 5.
142 Chapter 4. General Discussion
Test/Extinction. The test session on Day 2 employed identical parameters to that used in Experiment 5. This session was repeated on Day 3. No injections were administered prior to test/extinction sessions. To permit examination of any effects of
FG7142 on context conditioning, data from rats that froze >50% during the first 3 min of the test session was included in analyses.
Results and Discussion
Conditioning. Freezing to the tone during the conditioning session is shown in the left panel of Figure 15. Freezing during the first 3 min of context exposure was 0.6%
(SEM = 0.6) in group 0 mg/kg, 1.9% (SEM = 0.9) in group 1 mg/kg, and 0.6% (SEM =
0.6) in group 10 mg/kg. No pairwise between-groups comparisons of context freezing approached statistical significance (Fs (1, 20) < 2.0, ps > 0.05). Thus, there is no evidence that FG7142 causes unconditioned context freezing. Neither were there any between-group main effects of drug on CS freezing (Fs (1, 20) < 1, ps > 0.05). Freezing increased significantly between the first and second CS presentation (F (1, 20) = 59.8, p
< 0.05), but this increase did not interact significantly with pairwise between-group comparisons (Fs (1, 20) < 2.9, ps > 0.05).
Test/Extinction. Freezing to the tone during the test/extinction sessions, averaged over two-trial blocks, is shown in the right panel of Figure 15. Pairwise between-group comparisons of freezing during the first 3 min of the session did not reveal significant main effects (Fs (1, 20) < 2.4, ps > 0.05). Freezing during the first 3 min of the session declined between Day 2 and Day 3 from 8.0% (SEM = 4.9) to 2.3% (SEM = 1.7) in group 0 mg/kg, from 25.9% (SEM = 11.6) to 7.1% (SEM = 3.7) in group 1 mg/kg, and from 15.9% (SEM = 7.3) to 5.1% (SEM = 4.0) in group 10 mg/kg. This decline was statistically significant (F (1, 20) = 9.3, p < 0.05), showing that context fear extinguished between the two sessions, but this decline did not significantly interact
143 Chapter 4. General Discussion
Table 8. Design of Experiment 7
Group Conditioning Test
0 mg/kg 0 mg/kg FG7142 – Tone + Tone
1 mg/kg 1 mg/kg FG7142 – Tone + Tone
10 mg/kg 10 mg/kg FG7142 – Tone + Tone
Conditioning Test 100 100 0 mg/kg 1 mg/kg 75 75 10 mg/kg
50 50
25 25 PERCENT FREEZING 0 0 1 2 1 2 3 4 5 6 Trial Trial block
Figure 15. Mean CS freezing during each stage of Experiment 7. Freezing on test is shown averaged over blocks of 2 trials.
144 Chapter 4. General Discussion
with the between-group comparisons (Fs (1, 20) < 1.9, ps > 0.05). Thus there is no evidence that FG7142 affects contextual fear learning.
There were also no significant between-group main effects of FG7142 on freezing during tone presentations (Fs (1, 20) < 1, ps > 0.05). Extinction was apparent as freezing to the CS declined over the course of the 12 test presentations (F (1, 20) = 37.1, p <
0.05), but this decline did not interact significantly with any between-group comparisons (Fs (1, 20) < 1, ps > 0.05). Thus there is no evidence for any general effect of FG7142 on contextual or CS-specific fear acquisition or expression at short or long intervals. This adds to the findings in Experiments 4-6 that FG7142 does not significantly alter fear expression. Furthermore, it provides no evidence consistent with the possibility that the ability of the shock to condition fear was altered during Stage II of Experiment 4 in group Over – FG.
Experiments 8a and 8b
According to theories that treat contextual stimuli as occasion-setters for different memories attached to stimuli with mixed histories of reinforcement (e.g. Bouton, 1993), expression of the extinction memory is highly context-specific. Harris and Westbrook’s
(1998) finding that the effect of FG7142 on the expression of extinction was not additive with the effect of renewal is consistent with a role for GABA in mediating the context-specificity of extinction. One might, therefore, infer that if extinction and overexpectation of fear depend on a common GABAergic mechanism, they would therefore be similarly vulnerable to renewal. Rescorla (2007) has reported ABA, AAB, and ABC renewal in overexpectation of appetitive conditioning. To date there have been no published investigations of renewal from overexpectation in fear conditioning.
145 Chapter 4. General Discussion
Demonstrations of ABA renewal from overexpectation are technically difficult because several criteria must be independently assessed. First, the contexts used in a renewal design must be discriminable. Second, and more importantly, fear to the trained
CSs must show substantial generalisation from Stage I training in context A to Stage II training in context B. According to Equation 1, any generalisation decrement from
Stage I to Stage II will reduce the magnitiude of the negative prediction error generated by Stage II training and consequently reduce the magnitiude, or possibly even abolish, overexpectation. It was the intention of these experiments to provide independent evidence bearing on these criteria. Experiment 8a examined whether a shift in context between conditioning and test altered the expression of freezing to a target tone CS.
Experiment 8b examined the discriminability of the two contexts by testing generalisation of context conditioned fear between them.
Method
Subjects
Subjects were 32 male Wistar rats obtained, housed, and handled as in previous experiments. There were 16 rats in Experiment 8a and 16 rats in Experiment 8b.
Apparatus
Two sets of four experimental chambers were used. One set was the same as that used in Experiments 1 - 7, with the exception that the extractor fan was turned off during conditioning sessions to reduce background noise. In addition, four drops of almond essence were added to the bedding before to create a distinct odour. This is designated context 1. The other set of chambers, designated context 2, were 240 mm
(length) × 300 mm (width) × 210 mm (height). The top and rear walls of these chambers as well as the front hinged door were constructed of clear Perspex, and the end walls were made of stainless steel. The floor in each chamber consisted of stainless steel rods
146 Chapter 4. General Discussion
4 mm in diameter spaced 15 mm apart (centre to centre). Each chamber stood 2 cm above a tray of paper pellet bedding (Fibrecycle, Mudgeeraba, Australia). These four chambers were located individually within sound attenuating boxes that were painted white. The boxes were constantly illuminated by an 8 W white light bulb located in the rear wall of the box. Ventilation fans provided a constant background noise (67 dB [A scale]). In addition, 4 drops of rose essence were added to the bedding. Electronic programming and delivery of stimuli to these conditioning chambers and recording of rats’ behaviour was conducted in the same manner as in the other set of chambers.
Procedure
Experiment 8a. Conditioning. On the first day of experimental procedures, rats were placed in either context 1 (n = 8) or context 2 (n = 8) for 20 min. A tone, identical in its parameters to that used in previous experiments, was presented twice during this session. The first presentation occurred 6 min after the beginning of the session and the second 12 min and 30 s after the session commenced. Each presentation co-terminated with a 1 s, 0.5 mA shock.
Test. The following day, freezing to the tone was tested in a session identical to the test session used in Experiments 1 - 4. Test sessions were conducted in either the same (group same, n = 8) or different (group different, n = 8) context as that used for conditioning. Half of the rats conditioned in each context were in each of these groups. To enable observation of the contextual manipulation on fear of the context, data was not excluded from rats that froze >50%.
Experiment 8b. Conditioning. On Day 1, rats were placed into either context 1 (n =
8) or context 2 (n = 8), in which they were exposed to two 0.5 mA, 1 s shocks, which occurred 2 and 4 min after placement in the context. One min after the last shock, rats were removed and returned to their home cages.
147 Chapter 4. General Discussion
Test. The following day freezing was measured during a 4 min exposure to either the same context (group Same: n = 8) or to the other, novel context
(group Different: n = 8). Half of the rats conditioned in each context were assigned to each group.
Results and Discussion
Experiment 8a. Mean freezing during the first 3 min of context exposure during the test session and averaged across the 4 presentations of the tone is shown in Figure 16.
Context freezing was significantly higher in group Same than in group Different (F (1,
12) = 8.5, p < 0.05). This shows that the contexts were discriminable with no pre- exposure, because rats placed in the same context where they had experienced shock froze more than rats placed in a novel context. Rats did not significantly differ in context freezing depending on which of the two physical contexts they were conditioned (F (1, 12) = 1.9, p > 0.05) or tested in (F (1, 12) = 3.8, p > 0.05), providing no evidence that one physical context was more associable than another or that one context generally produced more freezing than another.
Freezing to the CS, averaged across all 4 tone presentations, was greater than context freezing (F (1, 12) = 77.2, p < 0.05). The interaction between the difference between context and CS freezing and group did not reach significance (F (1, 12) = 4.2, p =
0.063). Freezing to the tone was unaffected by group, conditioning context, or test context (Fs (1, 12) < 1, p > 0.05). These results show that tone freezing generalised across contexts, and was acquired and expressed equivalently in both contexts.
Experiment 8b. Mean freezing during the two 2 min epochs of the test session is shown for groups Same and Different in Figure 17. There was no effect of physical training or test context on freezing and no interaction between either of these contrasts and the within-subjects epochal contrast (Fs (1, 12) < 1, ps > 0.05). Rats appeared to
148 Chapter 4. General Discussion
100 Context CS 75
50
25
PERCENT FREEZING 0 Same Different Same Different
Figure 16. Mean freezing in Experiment 8a. Freezing is shown during both the first 3 min of context exposure and averaged over all 4 tone presentations.
100 Same Different 75
50
25
PERCENT FREEZING PERCENT 0 1 2
2-minute period
Figure 17. Mean freezing during test context exposure in Experiment 8b.
149 Chapter 4. General Discussion
acquire and express context fear equivalently in both physical contexts. There was a significant interaction between the group factor and the within-subjects comparison of freezing during the first vs. second 2 min period (F (1, 12) = 9.3, p < 0.05). As physical context of conditioning and test did not significantly affect freezing, group Same and group Different were each treated as single groups for the follow-up analyses of this interaction. During the first 2 min rats in both groups froze at similar levels (F (1, 14) <
1, p > 0.05) but during the second 2 min, group Same froze significantly more than group Different (F (1, 14) = 8.4, p < 0.05). These results confirm that the contexts are discriminable.
Experiment 9
The present experiment studied the context-specificity of overexpectation using an
ABA renewal design. Such renewal is a hallmark of extinction (Bouton, 2004) and has recently been observed in appetitive overexpectation by Rescorla (2007). The experiment was a single factor design. This design is shown in Table 9. Briefly, group
Control received Stage I but not Stage II training. Rats in this condition were tested in either the same or different context as Stage I training. Group AAA received Stage I,
Stage II, and test in context A. Group ABB received Stage I training in context A and
Stage II as well as test in context B. Finally, group ABA received Stage I training and test in context A but Stage II training in context B. Both AAA and ABB control conditions were used for two reasons. First, the AAA control is identical to the designs used previously for overexpectation in this thesis. However, this is a somewhat liberal control for renewal from overexpectation because it differs from the experimental condition, group ABA, in terms of location of Stage II training. Therefore an ABB control was also used to confirm the presence of overexpectation with a context change
150 Chapter 4. General Discussion
between Stage I and Stage II. If overexpectation is learned equivalently in contexts A and B, then groups AAA and ABB are equally useful as controls for renewal. For this reason, most analyses were conducted with groups AAA and ABB pooled as a single group Same (indicating that test was in the same context as Stage II training). This design is therefore a 3 group design: group Control, group Same, and group Different.
Method
Subjects and apparatus
Subjects were 63 male Wistar rats obtained, housed, and handled as in previous experiments. Apparatus was identical to that used in Experiments 8a and 8b.
Procedure
Experiment 9 adopted the conditioning and test procedures used in Experiment 3 and
4. The only difference being the location of the Stage II or test training (contexts 1 and 2 described previously). The same exclusion criteria used in previous overexpectation experiments were applied. Seven rats failed to freeze more than 30% to the tone on Day
3 and were excluded from further procedures. Data from a further 9 rats was excluded from analyses due to freezing levels >50% during the first 3 min of the test session. Of these excluded rats, 1 was in group Control, 3 in group Different, and 5 in group Same
(3 in group AAA and 2 in group ABB). Final group sizes were: Group Control (n = 18); group Different (n = 15); group Same (n = 14: group AAA n = 7; group ABB n = 7).
Results and Discussion
Stage I. Freezing to the tone and flashing light CSs during Days 3 and 4, respectively, of Stage I conditioning is shown in the left panel of Figure 18. CS freezing on these days was not significantly affected by physical conditioning context (F (1, 45)
< 1, p > 0.05). Context freezing on these days was, however, significantly higher in context 2 (mean = 25.2%, SEM = 4.6) than in context 1 (mean = 11.9%, SEM = 2.2) (F
151 Chapter 4. General Discussion
Table 9. Design of Experiment 9. Capital letters in parentheses refer to contexts.
Procedures for groups Control and Same are shown separately for each sub-grouping.
Group Stage I Stage II Test
Control (A) Flash +, Tone + ______(A) Tone
(A) Flash +, Tone + ______(B) Tone
Same (A) Flash +, Tone + (A) Flash/Tone+ (A) Tone
(A) Flash +, Tone + (B) Flash/Tone+ (B) Tone
Different (A) Flash +, Tone + (B) Flash/Tone+ (A) Tone
Stage I Stage II Test 100 100
75 75
50 50
25 25 PERCENT FREEZING 0 0 Tone Flash 1 2 Control Same Different CS Trial Group
Figure 18. Mean CS freezing during each Stage of Experiment 9. Left panel shows CS freezing on days 3 and 4 of Stage I conditioning. Freezing on test is shown averaged across all 4 tone presentations.
152 Chapter 4. General Discussion
(1, 45) = 7.6, p < 0.05). Thus, while Experiment 8 did not reveal any significant differences between freezing levels in context 1 and 2, the different procedures and larger numbers of rats used in the present experiment appear to have revealed subtle differences between these two contexts. Regardless, freezing to the CSs on these days was significantly higher than context freezing (F (1, 45) = 155.4, p <0.05), and this contrast did not interact significantly with conditioning context (F (1, 45) = 2.8, p >
0.05).
Stage II. Freezing to the context at the beginning of Stage II did not differ based on which physical context was used for Stage I or Stage II (Fs (1, 27) < 1, ps > 0.05). It did, however, differ based on whether or not there was a context-shift between Stage I and Stage II. Rats that received Stage II conditioning in the same context as Stage I froze more to the context (14.1%, SEM = 6.8) than rats that underwent a context shift
(1.7%, SEM = 1.1) (F (1, 27) = 8.9, p < 0.05). This suggests that a mild context-shock association was retrieved if the Stage II conditioning session occurred in the same context where rats had previously experienced shock, but not if this session occurred in a novel context. Thus, this adds further evidence that the two contexts used were discriminable. Freezing to the CS was higher than context freezing (F (1, 27) = 42.3, p <
0.05).
There was no evidence for summation during the first presentation of the compound stimulus (F (1, 27) = 1.3, p > 0.05). Freezing to the compound was not significantly affected by Stage I or Stage II physical context, nor by whether or not there was a context shift between these two stages (Fs (1, 27) < 2.7, ps > 0.05). The absence of an effect of contextual shift on freezing to the compound further confirms that CS freezing generalised across contexts. Furthermore, none of the contextual variables analysed interacted significantly with the change in freezing across the 2 compound presentations
153 Chapter 4. General Discussion
(Fs (1, 27) < 4.1, ps > 0.05). For this reason, the change in compound CS freezing during Stage II training was analysed with all rats treated as a single group. As in previous overexpectation experiments, there was a significant decline in freezing between the two compound presentations (F (1, 28) = 27.7, p < 0.05).
Test. Context freezing did not differ based on physical context of Stage I training or test (Fs (1, 45) < 1, ps > 0.05). Neither context (F (1, 16) < 1, p > 0.05) nor tone freezing (F (1, 16) = 2.8, p > 0.05) differed significantly within group Control as a function of whether rats were tested in context A or B so they were combined into a common control group for the purposes of further analyses. Context freezing was generally higher in groups receiving Stage II training than in group Control, which showed a mean freezing level of 5.2% (SEM = 1.2) (F (1, 44) = 5.2, p < 0.05).
Furthermore, context freezing in group Same (17.2%, SEM = 4.2) was significantly higher than in group Different (7.9%, SEM = 2.0) (F (1, 44) = 5.6, p < 0.05). However, further examination of the data, with group Same treated as two separate groups (Group
AAA and group ABB) suggests that this difference is due exclusively to freezing levels in group AAA. Group AAA showed significantly more context freezing (25.0%, SEM =
6.3) than group ABB (9.3 %, SEM = 4.1) and group Different (F (1, 43) = 14.4, p <
0.05), while the latter two groups did not differ significantly (F (1, 43) < 1, p > 0.05).
Freezing to the tone CS was higher than freezing to context (F (1, 44) = 117.3, p <
0.05).
Freezing to the tone CS on test, averaged across the four CS presentations, is shown in the right panel of Figure 18. CS freezing was not significantly influenced by either physical Stage I (F (1, 45) = 1.2, p > 0.05) or test context (F (1, 45) < 1, p > 0.05). There was no difference in CS freezing between group AAA and group ABB (F (1, 12) = 1.2, p > 0.05) so these groups were treated as a single group Same for the following
154 Chapter 4. General Discussion
analyses. There was evidence for overexpectation of fear because Groups Same and
Different displayed significantly less freezing on test than group Control (F (1, 44) =
8.5, p < 0.05). However there was no evidence for context-specificity of overexpectation because group Same did not differ significantly from group Different
(F (1, 44) < 1, p > 0.05).
It could be argued that the difference in context freezing between groups Same and
Different may have confounded detection of renewal in group Different. As group Same froze more to the context than group Different, summation of context and CS fear could have obscured actual differences in CS-specific fear between these two groups. For this reason, test data were re-analysed with rats from group AAA excluded from group
Same, because this abolished the difference between groups in context freezing. These data are depicted in Figure 19. The added benefit of conducting such an analysis is that it provides a direct test of whether overexpectation was learned in context B. Again, overexpectation was observed, because groups ABB and Different froze significantly less than group Control (F (1, 37) = 4.8, p < 0.05). Thus, overexpectation was learned in context B. Renewal was not detected, because there was no significant difference in freezing between group ABB and group Different (F (1, 37) < 1, p > 0.05).
The present experiment confirms the overexpectation of fear learning. However, unlike the results of Rescorla (2007), it did not observe renewal from such overexpectation. This discrepancy may have several causes. One is that the appetitive conditioning design used by Rescorla (2007) may generally produce more robust renewal than the fear conditioning design used in Experiment 9. The larger number of compound conditioning sessions and trials used by Rescorla (2007) may have better promoted the context acquiring ‘occasion-setting’ properties that controlled expression of performance during overexpectation. In the present design, in which overexpectation
155 Chapter 4. General Discussion
100
75
50
25
PERCENT FREEZINGPERCENT
0 Control ABB Different Group
Figure 19. Test data from Experiment 9, with group AAA excluded. Freezing is shown averaged across all 4 tone presentations.
156 Chapter 4. General Discussion
was learned in a small number of compound - shock trials in a single session, performance may have been less ‘linked’ to the context. Its expression may, therefore, not have been subject to contextual modulation, allowing it to generalise to the Stage I conditioning context. In any case, the failure to observe renewal in Experiment 9 is not conclusive evidence for the general absence of renewal after overexpectation in conditioned fear. It may be that with a different contextual manipulation, or different conditioning or testing parameters, renewal could be observed.
Experiment 10
Experiments 8 and 9 provided evidence that the contexts used were discriminable.
Such discriminability is necessary, but may not be sufficient, to produce renewal. The capacity of contexts to gain differential control over expression of fear after a decremental learning process may depend on additional factors. If the contexts used in
Experiment 9 were generally incapable of gaining such differential control, then this contextual manipulation would obviously be useless for investigation of renewal, rendering the results of Experiment 9 meaningless. One way to test this possibility is to examine whether renewal from extinction can occur using this contextual manipulation.
Experiment 10 examined this question by conditioning two groups of rats to fear a clicker in one of the two contexts used in Experiments 8 and 9. Group AAA then received extinction training in the conditioning context while Group ABA received this training in the other context. All rats were then tested with a single presentation of the clicker in the conditioning context. The design of this experiment is shown in Table 10.
157 Chapter 4. General Discussion
Method
Subjects and Apparatus
Subjects were 11 male Wistar rats obtained, housed and handled as in previous experiments. Apparatus was as described for Experiments 8 and 9.
Procedure
Pre-exposure. On Day 1, all rats were exposed to context 1 for 10 min.
Approximately 2 hours later they were exposed to context 2 for 10 min.
Conditioning. On Day 2, rats were placed in one of the two contexts where they were exposed to two presentations of a 30 s, 82 dB, 20 Hz clicker, both of which co- terminated with a 0.5 s, 0.6 mA shock. The first clicker presentation commenced 3 min after rats were placed in the conditioning context and the ISI was 3 min. Rats were removed from the conditioning chambers and returned to their home cage 1 min after the second shock presentation. On the same day, rats in group ABA were exposed to a 5 min exposure to the ‘B’ context (the non-conditioning context) during which no stimulus events were programmed. This additional exposure was designed to further enhance discrimination between the conditioning and extinction contexts in this group before extinction training commenced.
Extinction. On Days 3-5, all rats were subjected to a daily 18 min extinction session in which they were exposed to 8 non-reinforced presentations of the clicker, beginning 2 min after placement in the conditioning chambers, with a mean ISI of 90 s. For rats in group AAA, these sessions were conducted in the conditioning context, whereas in group ABA, these were conducted in the other context.
Test. On Day 5 all rats were placed in the chambers that had served as the conditioning context and exposed to a single 30 s presentation of the clicker, beginning
2 min after placement in the context.
158 Chapter 4. General Discussion
Data Analysis
Rats were scored for freezing every 4 s during the first 2 min of extinction and test sessions, and every 2 s during CS presentations. Data from one rat in group ABA was excluded due to freezing levels >50% during the first 2 min of the test session. Final groups sizes were: group AAA (n = 5); group ABA (n = 5). Each group included 2 rats conditioned in physical context 1 and 3 rats conditioned in physical context 2.
Results and Discussion
Extinction. Mean freezing to clicker presentations during extinction is shown in the left panel of Figure 20. In group AAA, freezing during the first 2 min context exposure was 66.0% (SEM = 11.8) on Day 3, 35.3% (SEM = 14.9) on Day 4, and 17.3% (SEM =
6.5) on Day 5. In group ABA, freezing during this period was 39.3% (SEM = 5.8) on
Day 3, 2.7% (SEM = 1.3) on Day 4, and 4.7% (SEM = 2.9) on Day 5. There was a significant difference between groups in context freezing during these days (F (1, 8) =
5.4, p < 0.05), confirming that the contexts were discriminable. There was also a significant linear decrease in context freezing across days (F (1, 8) = 53.9, p < 0.05), showing that context fear extinguished during extinction training. The interaction between the between-groups difference in context freezing and the within-subjects linear decline in freezing did not reach statistical significance (F (1, 8) = 1.5, p > 0.05).
The physical context in which extinction took place did not significantly affect context freezing (F (1, 8) = 1.6, p > 0.05) and did not interact significantly with the decrease in context freezing across days (F (1, 8) = 1.1, p > 0.05).
Freezing to the clicker also exhibited a linear decrease over the 24 extinction trials (F
(1, 8) = 39.9, p < 0.05). Analysis of CS freezing revealed no significant between-groups main effect or interaction between effect of group and the linear decrease in freezing over trials (Fs (1, 8) < 1, ps > 0.05). CS freezing was also unaffected by the physical
159 Chapter 4. General Discussion
Table 10. Design of Experiment 10.
Group Conditioning Extinction Test
AAA (A) Clicker + (A) Clicker - (A) Clicker
ABA (A) Clicker + (B) Clicker - (A) Clicker
Extinction Test 100 100 AAA ABA 75 75
50 50
25 25 PERCENT FREEZING 0 0 1 2 3 4 5 6 AAA ABA Trial block GROUP
Figure 20. Mean CS freezing during extinction and test in Experiment 10. Freezing during extinction training is shown averaged over 4-trial blocks.
160 Chapter 4. General Discussion
context in which extinction took place, and this contrast did not significantly interact with the linear decline across trials (Fs (1, 13) < 1, ps > 0.05).
Test. Freezing during the test CS presentation is shown in the right panel of Figure
20. Mean freezing during the first 2 minutes of the test session was 4.7% (SEM = 2.9) in group AAA and 18.7% (SEM = 8.1) in group ABA. This difference did not reach statistical significance (F (1, 8) = 2.7, p > 0.05). Context freezing was unaffected by physical context (F (1, 8) < 1, p > 0.05).
There was evidence for renewal as rats in group ABA froze more to the clicker than rats in group AAA (F (1, 8) = 12.0, p < 0.05). Physical context did not have a significant effect on freezing to the clicker (F (1, 8) < 1, p > 0.05). This evidence for renewal suggests that these contexts are able to acquire differential control over fear, as expressed through freezing, after extinction. This makes it more difficult to attribute the failure to observe renewal after overexpectation in Experiment 9 to an insufficiently powerful context manipulation.
161 Chapter 4. General Discussion
Chapter 4
General Discussion
This thesis studied the overexpectation of Pavlovian fear conditioning. The aim was to examine the effect of the benzodiazepine receptor partial inverse agonist FG7142 on the expression of fear overexpectation as well as the potential role of physical context in regulating expression of fear overexpectation. This thesis, therefore, studied potential similarities between the mechanisms regulating expression of fear after overexpectation and extinction training. Contemporary findings and theories regarding extinction suggest that it may involve multiple types of learning, which may engage distinct neurobiological substrates. It was argued in the Introduction that overexpectation is especially useful as a model of specific mechanisms of fear reduction caused by negative predictive error. The aim of this chapter is to relate the empirical findings reported in Chapter 3 to the theories of fear learning and fear extinction discussed in
Chapters 1 and 2.
1. Summary of empirical results and their theoretical implications
Experiment 1 showed that fear overexpectation could be detected when Stage II training consisted of 2 compound CS - US presentations. Further Stage II training caused an over-training effect resulting in reductions of overexpectation after 4 or 8 compound - US pairings. Experiment 2 showed that administration of the benzodiazepine partial inverse agonist FG7142 prior to test attenuated expression of overexpectation and so caused a significant increase in freezing in rats subjected to
Stage II training. Experiment 3 showed that the ability of FG7142 to attenuate expression of overexpectation was dose-dependent. Moreover, the dose-response 162 Chapter 4. General Discussion
properties of this effect of FG7142 were similar to those seen in other experimental preparations involving decrements in learned fear - extinction and infantile amnesia in rats. Experiment 4 suggests that the ability of FG7142 to attenuate expression of overexpectation was not due simply to state-dependent learning/ memory processes: rats injected with FG7142 before both Stage II and Test did not show overexpectation.
These results show that: 1) fear overexpectation, like fear extinction, is not simply the partial erasure of the original CS – US association; and, 2) fear that remains after overexpectation training is sensitive to modulation of the benzodiazepine binding site of the GABAA receptor. Taken together, these results suggest that a mask is imposed on expression of learned fear by overexpectation training and that this mask is mediated by
GABAA receptors.
A number of control experiments were completed to further understand the effects of
FG7142 on the expression of learned fear. FG7142 did not affect expression of freezing in the control group in Experiment 4 suggesting that its effect on overexpectation was not simply due to a general effect on expression of freezing. Experiment 5 and 7 further supported this conclusion. FG7142 did not significantly alter expression of freezing on test after conditioning with a low-intensity shock in Experiment 5 and did not induce unconditioned context freezing or affect CS freezing during a conditioning session in
Experiment 7. Furthermore, Experiment 6b showed that FG7142 did not alter freezing to a CS subjected to an associative blocking procedure. These results suggest that enhancement of freezing by FG7142 is specific to CSs which have been subjected to a procedure causing decrements in conditioned fear.
Further experiments investigated the possibility that fear overexpectation would show context-mediated renewal. Experiments 8a and 8b showed that the context manipulation caused greater freezing to a shocked context than to a non-shocked
163 Chapter 4. General Discussion
context. This suggested that the contexts were discriminable. Experiment 8a also showed that freezing to a CS paired with shock in one context generalised to the other context. Experiment 10 showed that the contextual manipulation allowed ABA renewal after extinction of a feared clicker CS. However, Experiment 9 yielded no evidence for
ABA renewal from overexpectation. A significant overexpectation effect learned in the
‘B’ context was expressed regardless of whether test was conducted in the Stage II or
Stage I training context.
The findings from this thesis can be summarised as follows:
1. The detection of overexpectation using freezing as a measure of conditioned fear
(McNally et al., 2004a) has been replicated, and adds to evidence that this effect is common across numerous measures of associative learning. This confirms the more general hypothesis that simple temporal contiguity of a CS and US is insufficient for causing increments in fear. This is consistent with the action of prediction error on the formation of CS - US associations, but is also possibly consistent with the action of a comparator process in the expression of a learned CS - US association. The over- training of overexpectation detected in Experiment 1 is, to the best of the author’s knowledge, the first such demonstration.
2. FG7142 causes recovery of a freezing CR that has been attenuated by fear overexpectation. This is not simply caused by FG7142 augmenting the freezing response. This suggests that common mechanisms contribute to expression of fear after overexpectation and extinction training. In particular, it suggests that, like fear extinction, the portion of the CS - US association not expressed after fear overexpectation has not been erased, but is instead masked in a GABAA receptor-
164 Chapter 4. General Discussion
dependent manner. This evidence that the CS – US association is not simply erased by overexpectation training appears inconsistent with the predictions of the Rescorla-
Wagner model in Equation 1.
3. FG7142 appears only to alter expression of fear to a CS subjected to a decremental training procedure. FG7142 did not affect expression of fear to an associatively blocked stimulus or a stimulus that has been simply conditioned and then tested. This dissociation between blocking and simple fear expression on the one hand, and overexpectation and extinction on the other, is consistent with the hypothesis that
GABAergic masking of learned fear is a specific consequence of negative predictive error.
4. There was no evidence for ABA renewal from overexpectation, despite evidence that the contexts used were discriminable and could produce renewal from extinction.
Overexpectation was learned in a context other than that used for Stage I training, but was still expressed when rats were tested in the Stage I training context. While this could, in light of evidence for renewal from appetitive overexpectation (Rescorla,
2007), be seen as evidence for a dissociation between appetitive and aversive overexpectation, and also between fear extinction and overexpectation, it is premature to draw such conclusions based on a single result.
5. It should be noted that overexpectation was consistently observed in Experiments 1-4 and in Experiment 9, despite evidence for summation during Stage II training only being observed in Experiment 1. This suggests both that excitatory summation of
165 Chapter 4. General Discussion
freezing is not a robust phenomenon, and that overexpectation can occur independently of evidence for summation in performance.
2. The over-training of overexpectation
As mentioned in Chapter 1, overexpectation has been observed in fear conditioning in rats (Rescorla, 1970; Kremer, 1978; Blaisdell et al., 2001; McNally et al., 2004a), appetitive conditioning in rats (Lattal & Nakajima, 1998; Rescorla, 1999; 2006; 2007) and pigeons (Khallad & Moore, 1996), eye-blink conditioning in rabbits (Kehoe &
White, 2004), and causal judgement learning in humans (Collins & Shanks, 2006).
Thus, the demonstration of overexpectation in Experiment 1, on its own, is not a novel result. The over-training of overexpectation is, however, a novel result. It was also unexpected in light of Kremer’s (1978) report that overexpectation was most robust with greater numbers of compound - shock pairings during Stage II training.
Furthermore, over-training of overexpectation is apparently inconsistent with the
Rescorla-Wagner model. According to Equation 1, if the combined associative strengths of the two CSs presented during Stage II exceed the λ value of the US at the beginning of Stage II, then each compound - shock pairing should cause a decline in CS associative strength until ∑V = λ. At this point, further compound - shock pairings should cause no change in associative strengths.
Over-training of overexpectation is allowed for by the extended comparator model proposed by Stout and Miller (2007). However, it is only clearly predicted if the test context is different to either the Stage I or II training contexts. This model explains such effects as occurring due to second-order comparator processes, which attenuate the strength of the primary comparator process producing the overexpectation effect. As explained previously, this model proposes that if a target stimulus ‘X’ (e.g. the tone CS
166 Chapter 4. General Discussion
in this thesis), which directly activates an internal US representation, also activates an internal representation of a stimulus ‘A’ (e.g. the flashing light CS), which is also associated with the same US, performace of a CR may be modulated by a comparator process. This process, referred to as a first-order comparator process, results from the indirect activation of the US representation by the A representation, and a comparison between the strength of the directly- and indirectly-activated US representations.
However, Stout and Miller (2007) also allow each of the associative links that lead to the activation of the indirect US representation to be subject to second-order comparator processes. A graphical depiction of this theory is included in Figure 21. If X or A are associated with a third stimulus ‘B’ (i.e. the context in these experiments), which is also associated with either A or the US, X may either directly or indirectly activate a representation of B. This, in turn, may attenuate the first-order comparator process and thereby increase the CR. One example of a stimulus which Stout and Miller (2007) propose can act as a second-order comparator is the context that serves as a background to conditioning of an AX compound. Context-shock associations may develop more slowly than the tone-shock and flash-shock associations if the temporal correlation between context presentation and US-delivery is weaker than for discrete CSs. Thus, where a conditioning context serves as a second-order comparator stimulus, the second- order comparator process may emerge more slowly than primary comparator effects.
Indeed there is evidence that over-training of overshadowing can result from acquisition of context conditioning. Stout, Arcediano, Escobar, and Miller (2003) conditioned rats with 4 pairings of an auditory test CS with shock in compound with a more salient stimulus. When tested in a second context, rats showed less fear than rats conditioned with the test CS alone. Over-shadowing was not seen for a CS that received similar compound - shock pairings over 36 trials. Consistent with the hypothesis that
167 Chapter 4. General Discussion
Figure 21. An example of a second-order comparator process. In this case, associations between A and B and between B and the US (‘O’ in the figure) act on the A – US link to modulate the strength of indirect activation of the US representation. This model also proposes that a similar second-order comparator process may act upon the X – A associative link, modulating the activation of the internal representation of A by physical presentation of X, if presentation of X excites a representation of a stimulus that is also associated with A. Image obtained from Stout, C. C. and Miller, R. R.
(2007). Sometimes-competing retrieval (SOCR): A formalization of the comparator hypothesis. Psychological Review, 114, 759-783.
168 Chapter 4. General Discussion
this loss of over-shadowing was the result of context-shock associations, post- conditioning extinction of the conditioning context restored over-shadowing in rats conditioned with 36 compound - shock pairings.
However, the model advanced by Stout and Miller (2007) is only intended to explain performance to a CS presented alone, in the absence of putative comparator stimuli. In such situations, it is the associative activation of the of the first- and second-order comparator stimuli that modulates performance, rather than the physical presence of these stimuli. Thus, Stout et al. (2003) tested rats for the over-shadowing effect in a different context than that used for conditioning. When the test CS was presented, neither the comparator CS nor the second-order comparator stimulus (the conditioning context) were physically present. However, in Experiment 1 reported in this thesis, the same context was used for every stage of conditioning. Thus, on test, the conditioning context was physically present. In these circumstances, it is not clear whether the context could have acted as a second-order comparator, according to the extended comparator model. Only first-order comparator effects, mediated by activation of the internal representation of the flashing light CS, are clearly predicted by this model.
Thus, it is unclear how this model could explain the over-training of overexpectation observed in Experiment 1.
A second possible explanation for over-training of overexpectation appeals to the formation of a within compound tone – flash association during Stage II training. The more Stage II training trials, the stronger this association is. Given sufficient Stage II training, test presentations of the tone may activate both the tone – shock and tone – flash associations. The flashing light should at least partly retain the ability to elicit fear after overexpectation. Thus, the tone – flash association may elicit second-order conditioned fear which may summate with remaining fear elicited by the tone. Thus,
169 Chapter 4. General Discussion
like the comparator model, this explanation appeals to the formation of within- compound associations during Stage II training. However, unlike in the comparator model, these associations have an excitatory effect on fear performance, attenuating performance of overexpectation rather than causing it. According to this explanation, extinction training of the flashing light between Stage II training and test may attenuate or abolish over-training of overexpectation.
A third possible explanation of the over-training of overexpectation is that, over the course of repeated compound – shock presentations, configural learning processes replaced elemental learning processes. Such a mechanism requires the assumption that, given the parameters of Experiment 1, a configural representation of the tone – flash compound developed gradually over the course of Stage II training. Thus, during the early trials of Stage II, the tone and flashing light are processed as separate elements, and associative learning proceeds according to Equation 1, generating overexpectation.
However, as the configural representation forms over the course of compound presentations, a new association forms between this configural representation and the shock, independent of the pre-existing elemental associations.
Because overexpectation is generated during the first few Stage II pairings, test presentations of the tone CS only produce weak fear excitation through the direct tone- shock association. However, according to the configural learning model proposed by
Pearce (2002), a CS may elicit a response that has been conditioned to a configural stimulus to the extent that it can excite the configural representation. The degree to which it can do so is a function of its similarity to the configural stimulus. As the tone shares some features in common with the tone-flash compound, it should therefore be able to partially activate this configural representation. This may generate additional fear through the configural representation’s association with shock, and this additional
170 Chapter 4. General Discussion
fear may occlude the overexpectation effect. If such mechanisms contribute to the over- training of overexpectation, this would imply that treatments that enhance discrimination between the tone and flashing light should reduce and/or delay this effect. One such treatments is pre-exposure to the CSs. It is worth emphasising that such an explanation is inconsistent with either the Rescorla-Wagner or the Pearce (2002) learning rules. Rather it appeals to a switch from elemental to configural learning processes across extended Stage II training. Delamater, Sosa, and Katz (1999) have similarly suggested that contributions of both elemental and configural processes, with elemental processes being engaged earlier than connfigural processes, could explain effects they observed of prior discrimination training on summation and negative patterning learning in appetitive conditioning in rats.
A fourth explanation for the over-training effect may be that multiple shock presentations caused gradual sensitisation to the shock. If the λ value of a US is allowed to increase with sensitisation, this could allow over-training of overexpectation to be explained within the framework of Equation 1. At the beginning of Stage II, the λ value of the shock would be the same as during Stage I, resulting in overexpectation.
However, as the λ value increases over successive shock presentations, new excitatory fear learning may reverse this overexpectation. If such processes contributed to over- training of overexpectation, then simply continuing tone-shock pairings at the same timing parameters as used in Stage II conditioning may also cause shock sensitisation and increments in tone fear. In this case, a group exposed to 8 compound-shock pairings during Stage II conditioning may still show reduced fear relative to a control group exposed to 8 tone-shock pairings during this stage, despite not showing an overexpectation effect relative to a control group receiving no Stage II training of any sort.
171 Chapter 4. General Discussion
Whichever mechanism is chosen to explain over-training of overexpectation, the question arises as to why Kremer (1978) did not obersve such an effect. There are numerous parametric differences between Experiment 1 and Kremer (1978). For example, in Experiment 1, Stage II training was a single session, utilising an inter-shock interval of 5 min. Kremer (1978) used a procedure in which the inter-shock interval was
27 minutes. Moreover, in groups receiving 8 or 16 compound - shock pairings, Stage II training was spaced across 2 or 4 sessions respectively. Furthermore, Kremer (1978) conducted conditioning with a briefer (0.5 s) but more intense (1 mA) shock than that used in Experiment 1, and this conditioning was superimposed on operant bar-pressing.
Moreover, conditioned fear was assessed by freezing in the present experiments and by conditioned suppression in Kremer (1978). Further parametric investigations may clarify the conditions under which over-training of overexpectation may occur and thus help to elucidate the mechanisms by which this effect occurs.
3. Theoretical implications of the effect of FG7142 on expression of overexpectation.
While overexpectation was first reported nearly 40 years ago (Rescorla, 1970), it is only in the past decade that evidence has emerged to suggest that it does not simply involve partial erasure of the CS - US association. Rescorla (1999) first demonstrated this by showing that Pavlovian-instrumental transfer was equally strong in the presence of a CS previously subjected to overexpectation training as compared to an excitatory
CS not previously subjected to such training. Instrumental conditioning was conducted so that one response was rewarded with one outcome, and a second response rewarded with a second outcome. Pavlovian conditioning was then conducted in which one visual
CS and one auditory CS were paired with the first outcome, and a second visual and
172 Chapter 4. General Discussion
second auditory CS were paired with the second outcome. This was followed by pairings of a compound of one visual and one auditory stimulus with one of the two outcomes. Though this procedure resulted in reduced Pavlovian responding to the auditory CS involved in compound conditioning relative to the other auditory CS, each
CS equally and selectively augmented performance of the operant response associated with the specific outcome that it was associated with.
While this provided evidence for the survival of CS - US associations after overexpectation training, the first evidence that the Pavlovian response itself could recover after overexpectation came from Blaisdell et al. (2001). If, between Stage II and test, extinction was conducted with one of the two CSs that had been presented in compound during Stage II, less overexpectation to the other CS was observed. Further conditioning of one CS between Stage II training and test, in which the temporal relationship to the US was different to that used in Stage I and II training, had a similar effect. Following this, Rescorla (2006; 2007) detected spontaneous recovery and context-mediated renewal from overexpectation. The demonstration in this thesis that
FG7142 blocks expression of fear overexpectation, as measured by freezing, adds to this accumulating evidence that the reduction in a CR after overexpectation does not just reflect partial erasure of the original CS - US association. This body of evidence stands in contrast to the apparent predictions of the Rescorla-Wagner model.
Importantly, it raises the question of how fear is masked after overexpectation training.
The remainder of this section considers possible answers to this question.
3.1. A role for a comparator process?
Blaisdell et al. (2001) interpret their recovery from overexpectation in terms of the comparator hypothesis. Blaisdell et al. (2001) support this argument by presenting
173 Chapter 4. General Discussion
evidence that change from trace to delay conditioning of one CS after compound conditioning causes recovery of lick suppression to the other CS. This result is difficult to reconcile with any current model of error-correction, and is thus uniquely predicted by the comparator hypothesis. In the framework of the comparator hypothesis, the effect of FG7142 on fear overexpectation suggests that the comparator process is sensitive to modulation of the benzodiazepine site. Because the comparator hypothesis claims that the mechanisms for expression of overexpectation and blocking are similar, it implies that expression of blocking should also be vulnerable to FG7142. Experiment 6b found no evidence that FG7142 attenuated expression of blocking. Indeed, blocking was robustly expressed in the presence of FG7142.
This finding argues strongly against an interpretation of the effects of FG7142 in terms of a comparator process. But it must be emphasised that different parameters were used to produce blocking and overexpectation. All overexpectation and blocking experiments involving FG7142 involved the same number of compound - shock pairings with similar timing parameters. Thus, the tone-flash association would be expected to reach a similar strength in all these experiments. However, in the overexpectation experiments, the tone was paired with shock either 5 (Experiment 2) or
4 (Experiments 3 and 4) times, while in Experiment 6b, it was only paired with shock twice. This could possibly render the direct tone - shock association activated on test weaker in the blocking experiment than in the overexpectation experiments.
Furthermore, the flashing light was paired with shock 4 - 5 times in overexpectation experiments and 14 times in the blocking experiment. Thus, on test, indirect activation of the US representation, via the flash-shock association, may have been stronger in the blocking experiment than in the overexpectation experiments. These differences might contribute to differences in sensitivity to FG7142.
174 Chapter 4. General Discussion
Further experimentation to compare blocking and overexpectation within a single experiment would provide better clarification of this issue. An ideal experiment to address this may involve a within-subjects design in which stimuli A, B, and C are paired with shock during Stage I, and stimulus D is not. During Stage II, compounds
AB and CD would be paired with shock. This should result in overexpectation learning accruing to B, and blocking of D. Parameters could be arranged to either ensure that both compounds were paired with an equal number of shocks or that each target stimulus was paired with an equal number of total shocks. Ideally, parameters could be arranged to ensure equivalent freezing to both B and D on test in the absence of
FG7142. If, in such circumstances, FG7142 selectively increased freezing to B relative to a vehicle control group, this would provide much stronger support for the hypothesis that decrement of the CR is the specific result of negative predictive error.
3.2. How is fear masked after overexpectation?
The present experiments add weight to the contention that overexpectation shares common features with extinction, thereby adding to a body of evidence consistent with the hypothesis that they share similar mechanisms. Viewed in combination, the studies reported by McNally and Westbrook (2003) and McNally et al. (2004a) suggest that acquisition of both extinction and overexpectation rely on opioid receptor activation.
This is consistent with their hypothesis that opioid receptor activation signals the ‘-ΣV’ component of the parenthetical term in Equation 1. Furthermore, taken together, the results reported by Harris and Westbrook (1998) and those presented in this thesis suggests that the negative prediction error during extinction and overexpectation supports a process that masks fear upon later presentations of the CS in a GABAA receptor-dependent manner.
175 Chapter 4. General Discussion
In this thesis, and elsewhere (e.g. Barad, 2006; Myers et al., 2006, Myers & Davis,
2007) it has been suggested that extinction may be the result of multiple mechanisms.
These may include erasure, new excitatory learning (CS – No US association), inhibitory learning, CS habituation, and other mechanisms. Theoretical discussions of extinction have commonly invoked the formation of a “no-US” memory. This implies that the process that mediates expression of extinction is the specific consequence of the absence of reinforcement. Evidence that a similar process mediates expression of overexpectation, however, suggests that such a process can occur even in the presence of continued reinforcement of the CS. This does not exclude the involvement of a ‘no-
US’ memory as a possible component of extinction, but it does challenge the necessity of such a mechanism for decrements in conditioned fear.
If extinction is indeed the result of multiple mechanisms, it follows that different components of extinction may be under different degrees of contextual control. It is, therefore, of some interest that no evidence of renewal was observed in Experiment 9.
This suggests that any mechanism common to both fear extinction and fear overexpectation may be context-independent, and that the context-specificity of fear extinction may be the result of other, extinction-specific mechanisms. However, it would be premature to draw this conclusion on the basis of a single negative result. As reviewed earlier, Rescorla (2007) reported context-mediated renewal from overexpectation of appetitive conditioning. While it cannot be assumed that findings in appetitive preparations automatically apply to aversive preparations, further experimentation is necessary to examine the possible context-specificity of fear overexpectation. Such further work may well benefit from treatments designed to enhance the occasion-setting properties of context.
176 Chapter 4. General Discussion
The conclusion that overexpectation involves the imposition of a mask on fear raises the question of how this mask actually inhibits fear. Such inhibition could occur at the level of CS processing, inhibition of the US representation, inhibition of a central motivational state (in this case, fear), inhibition of specific behavioural responses, and perhaps other forms of “inhibition”.
According to Wagner (1981) inhibitory associations form between stimulus nodes activated to A1 and other stimulus nodes simultaneously activated to A2. The inhibitory memory process described by this model can therefore be conceived as the ability of CS nodes to reduce the likelihood of activation of US nodes to the A2 state. During an initial conditioning trial, both CS and US nodes are activated to A1 causing strong excitatory learning. Further presentations of the CS cause some US nodes to be activated to A2, meaning that further CS - US pairings result in a mixture of excitatory and inhibitory learning. However, if two CSs have excitatory associations with two different sub-sets of US nodes, then it is possible that compound presentation of these
CSs will excite more US nodes to the A2 state during an overexpectation trial than either would alone. This may allow inhibitory learning to exceed excitatory learning on any such trial.
Wagner’s (1981) theory essentially proposes that the inhibitory memory formed as a result of overexpectation consists of inhibition of the US representation. However, the findings reported by Rescorla (1999) present problems for this view of overexpectation.
The preservation of outcome-specific Pavlovian-instrumental transfer after overexpectation (or after extinction; Delamater, 1996) suggests that the CS remains able to excite the US representation even when Pavlovian responses to the CS itself suggest otherwise. These results have been interpreted by Rescorla (1993; 2006) to suggest that the learning resulting from extinction and overexpectation is the development of an
177 Chapter 4. General Discussion
inhibitory CS - response association, rather than an inhibitory CS - US association.
Furthermore, Rescorla (1999) found that overexpectation occurred when two CSs separately paired with two different, but motivationally similar, rewards, were presented in compound and paired with one of the two rewards. This suggests that prediction related to the general motivational qualities of the US, rather than its specific sensory characteristics, generates this inhibitory CS - response association.
The conditioning of opponent processes, as described by Schull (1979), provides a model for inhibitory CS-response association formation. Schull’s model is itself an extension of the opponent-process theory advanced by Solomon and Corbit (1974), which proposes that every motivational state produced by a biologically significant stimulus (called an ‘A process’) is countered by an opponent, endogenous ‘B’ process, designed to restore motivational equilibrium. According to this theory, with repeated experience of a particular A process, the opponent B process is strengthened in terms of magnitude and speed of onset after stimulus presentation. The motivational or affective state of an animal is the sum of these two opposing processes. Thus, when the A process is stronger than the B process, an animal experiences the ‘A state’, whereas a stronger B process may cause the animal to experience the ‘B state’.
Schull’s (1979) extension to this model was to suppose that the B process is a CR that can be elicited by CSs predicting such biologically significant stimuli. Thus, an aspect of the CR is an inhibition of the A state. When no US is presented, or when the
US presented produces an A process weaker than the B process elicited by CSs present
(as may occur in overexpectation training), the B state is experienced. Essentially, this means that the outcome of such a trial is a CS – ‘B state’ pairing. Schull (1979) argues that when a CS is paired with a B state, its ability to elicit CRs decreases and/or its ability to inhibit these CRs increases.
178 Chapter 4. General Discussion
Thus, this model suggests that the error-correction that occurs during overexpectation or extinction is generated by erroneous prediction of the motivational qualities of the outcome, rather than prediction of a specific US. In other words, overexpectation and extinction involve learning about the motivational state, rather than the specific stimulus, that follows a CS presentation. This learning, in turn, alters the ability of later CS presentations to excite the conditioned response (including, most importantly, the motivational response), rather than the US memory. While Schull is agnostic as to whether this response loss is due to erasure or to a new inhibitory process,
Rescorla (1993; 1999; 2006) proposes the latter. Thus, this view of overexpectation and extinction holds that the mask imposed on fear after overexpectation and extinction training involves inhibition of the aversive motivational response. FG7142 is not disinhibiting the US representation associated with the CS, but is instead disinhibiting the fear response. This distinction is important for predicting more general effects of
FG7142 on inhibited CRs, because it allows the hypothesis that the disinhibition produced by FG7142 may be specific to fear, rather than generally applicable to CS -
US associations. Indeed, this distinction raises the possibility that FG7142 may have opposing effects on expression of overexpectation and extinction of Pavlovian appetitive conditioning to those observed with aversive conditioning.
Several theorists have followed Konorski (1967) and proposed that the aversive and appetitive motivational systems are mutually inhibitory, meaning that a stimulus that excites one system is analogous to an inhibitor of the other, and an inhibitor of one system is essentially an excitor of the other (for a review of this theory and the empirical evidence for it, see Dickinson & Pearce, 1977). Thus, if FG7142 removes the inhibition imposed on the aversive motivational system by extinction or overexpectation training, it may be expected to have the opposite effect on the appetitive system. This leads to the
179 Chapter 4. General Discussion
hypothesis that FG7142 may actually enhance expression of appetitive extinction and overexpectation. If, however, FG7142 generally disinhibits CS – outcome associations, it should interfere with inhibition of Pavlovian CRs regardless of whether they are appetitive, aversive, or motivationally neutral.
To date, only one study has been published exploring the effects of FG7142 on appetitive Pavlovian responding. Delamater et al. (In Press) conditioned rats to associate a tone with delivery of food in one context, and a light with food in a second context. Extinction of each 15 s CS was then conducted in the context in which it had not been conditioned. Testing of the ability of each CS to elicit entry into the food- delivery magazine during both its presentation and the 20 s post-CS period was conducted in one context a day after the end of extinction training, and in the other context 4 days later. In rats injected with 0 or 2.5 mg/kg FG7142 15 min before each test session, responding to each CS was higher in its conditioning context than in its extinction context during both the CS and post-CS periods, demonstrating ABA vs.
ABB renewal. In rats injected with 5 mg/kg FG7142, this renewal was evident during the CS period, but not during the post-CS period. 10 mg/kg abolished renewal during both periods by selectively reducing the CR in the conditioning, but not extinction, context.
Delamater et al. (In Press) similarly showed that 10 mg/kg FG7142 abolished spontaneous recovery. Rats were conditioned to associate one CS (either a tone or white noise) with one US (either a food pellet or sucrose solution) over 3 days of training.
Extinction training was conducted with this CS on the 4th day. Over the next 4 days, conditioning and extinction training was conducted with the other CS and US in the same context. After this extinction session, rats were immediately injected with the drug or vehicle and tested for responding to both CSs 10 min later. Rats injected with vehicle
180 Chapter 4. General Discussion
showed more responding to the CS subjected to extinction training 4 days prior to test than to the CS subjected to this training 10 min prior to test, demonstrating spontaneous recovery. Rats injected with FG7142, however, showed similarly low responding to both CSs. FG7142 was not found to have any significant effect on expression of latent inhibition or its contextual modulation, acquisition of appetitive responding, or expression and within-session extinction of the response after a simple conditioning procedure.
These findings suggest that in appetitive conditioning, unlike aversive conditioning,
FG7142 strengthens, rather than weakens, expression of extinction. These effects of
FG7142 could be explained within the context of appetitive-aversive interactions if it is supposed that a GABAergic mechanism, sensitive to modulation by activity at the benzodiazepine receptor site, is involved specifically in the inhibition of the aversive system by the appetitive system. In simple appetitive or aversive acquisition, only one system comes to be activated by the CS. Thus, interaction between these two systems is not sufficiently involved for FG7142 to have an effect on acquisition or expression of a
CR in either appetitive (Delamater et al, In Press) or aversive (Experiment 7) conditioning. Aversive extinction or overexpectation allows the appetitive system to be sufficiently excited in the presence of the CS to at least partly inhibit activation of the aversive system. FG7142 removes this inhibition, restoring the aversive CR.
Appetitive extinction, according to this theory, involves activity in the aversive system, which comes to inhibit the appetitive response. As FG7142 does not prevent aversive inhibition of the appetitive system, FG7142 does not modulate CR expression when testing is conducted at a short interval after extinction and in the physical extinction context. However, in renewal and spontaneous recovery, the appetitive system may be more strongly activated relative to the aversive system, and thus able to
181 Chapter 4. General Discussion
inhibit activation of the aversive system. If FG7142 specifically interferes with this inhibition, it should restore the dominance of the aversive system, reinstating expression of extinction, as observed by Delamater et al. (In Press). It would be of interest to examine whether FG7142 modulates responding after appetitive overexpectation and if these effects are similarly dependent on context and passage of time. It would also be of theoretical interest to examine whether the effects of FG7142 extend to expression of aversive and appetitive devaluation, as this response decrement is generated by training that occurs in the absence of CS presentation and thus, according current theories, is due to alterations in US value rather than error-correction.
4. What are the neuroanatomical substrates for expression of overexpectation?
According to the ‘orthodox’ model of the amygdala’s role in fear conditioning, the
LA is critical for storage of a CS – US association, while the CeA is critical for output of the fear CR. Within the context of this model, the IL – ITC pathway is particularly appealing as a hypothetical substrate for expression of an inhibitory CS – response association. According to theories of this pathway’s role in extinction, its activation prevents CS-induced LA activation from disinhibiting CeM output neurons that mediate fear responses. Considering data suggesting that extinction involves changes in neural transmission and GABAergic tone in the BLA (Quirk et al.,1995; Rogan et al., 1997;
Hobin et al., 2003; Chhatwal, Myers, et al., 2005; Heldt & Ressler, 2007), and that vmPFC activity can modulate activity of BLA neurons (Rosenkranz & Grace, 2002;
Rosenkranz et al., 2003), such a model is probably too simplistic. Nevertheless, it is useful inasmuch as it leads to a number of testable hypotheses worthy of investigation.
If, as this thesis suggests, fear extinction and fear overexpectation share common mechanisms, it is worth investigating whether neural pathways implicated in inhibiting
182 Chapter 4. General Discussion
fear after extinction also mediate expression of fear overexpectation. If so, it would be expected that the burst firing seen after extinction training in IL (Burgos-Robles et al.,
2007) should also be seen in after overexpectation. Similarly the potentiation of IL unit firing to an extinguished CS (Milad & Quirk, 2002) should also be seen to a CS presented alone after overexpectation training. Overexpectation may also be vulnerable to treatments that impair consolidation of extinction, such as infusion of NMDA receptor antagonists (Burgos-Robles et al., 2007; Sotres-Bayon, Diaz-Mataix, Bush, &
Ledoux, in press) or a MAPK inhibitor (Hugues et al., 2004; 2006) into the mPFC immediately after training or infusion of anisomycin into the vmPFC before training
(Santini et al., 2004). Finally, inactivation of the IL, or lesions that disconnect it from the amygdala before test may prevent expression of overexpectation.
As mentioned previously, changes in BLA GABAA receptor expression and neural activity have been observed after extinction. This suggests that a model that relies on all extinction-related plasticity occurring in the vmPFC, and all extinction expression being mediated through inhibition of the CeA, is overly simplistic. Indeed, different sub- nuclei of the BLA, or different populations of neurons within the BLA, could participate in various functions described above (storage of the CS - US association, expression of fear CRs, and expression of an inhibitory CS-response association). This may be mediated by projections from the vmPFC, or may occur independently of such modulation. Thus, in addition to the hypotheses suggested above for testing, further hypotheses focused on a role for the BLA in fear overexpectation are worthy of investigation.
Investigation of a possible role for the BLA in fear overexpectation could include examination of whether acquisition of fear overexpectation, like fear extinction (Falls et al., 1992; Lee & Kim, 1998; Lin, Yeh, Lu, & Gean, 2003; Mao et al., 2006; Sotres-
183 Chapter 4. General Discussion
Bayon et al., 2007), requires NMDA receptor function in the BLA. Furthermore, consolidation of overexpectation may require activation of at least some of the same intra-cellular signalling cascades in the BLA that are involved in extinction (Lu et al.,
2001; Lin, Yeh, Leu, et al., 2003; Lin, Yeh, Lu, & Gean, 2003; Herry et al., 2006; Mao et al., 2006). Finally, the implication of GABAA receptors in expression of overexpectation suggested by this thesis raises the hypothesis that changes in the expression of this receptor and/or its level of activation by GABA may be involved in the consolidation of overexpectation. Examination of whether changes in H3- flunitrazepam binding, gephyrin mRNA and protein, and GAD67 mRNA expression seen after extinction (Chhatwal, Myers, et al., 2005, Heldt & Ressler, 2007) are also seen in overexpectation, could examine this question.
5. The relationship between the pharmacology of FG7142 and disinhibition of fear.
As a drug that negatively modulates the sensitivity of GABAA receptors to GABA, there are several mechanisms and locations of action by which FG7142 may interfere with masking of conditioned fear. The putative role for BLA GABAA receptors in expression of fear extinction (Chhatwal, Myers, et al., 2005; Barad et al., 2006) suggests that FG7142 may act at these receptors to disinhibit fear CRs. Alternatively, the role for of GABAergic ITC neurons in expression of extinction suggests that
FG7142 action at the targets of these neurons project may modulate expression of fear after overexpectation. Because FG7142 and other currently available benzodiazepine inverse agonists have very low solubility, it is difficult to selectively administer these chemicals to specific brain structures through microinfusion. However, localised microinfusion of a benzodiazepine antagonist, such as flumazenil, combined with systemic administration of FG7142 could be used to investigate the relative importance
184 Chapter 4. General Discussion
of different populations of GABAA receptors in expression of extinction and overexpectation.
The assumption that FG7142 impairs expression of extinction and overexpectation by its action at GABAA receptors leads to a further question: Are specific subtypes of
GABAA receptors involved in the expression of overexpectation? Several pieces of evidence suggest that different GABAA receptor subtypes are specifically involved in different extinction-related (and other) functions. For example, Heldt and Ressler (2007) found that changes in GABAA receptor subunit expression in the CeA after extinction were only seen for the α2 subunit.
More generally, the specific α subunit composition of GABAA receptors can lead to distinct effects of benzodiazepine site modulation. Point mutation of the α1 subunit in mice to render it insensitive to benzodiazepine site ligands has been found to abolish the sedative and amnesic effects of diazepam, and attenuate its anti-convulsant effect, while leaving intact its effects on muscle relaxation, motor coordination, and measures of anxiety (Rudolph et al., 1999; McKernan et al., 2000). Point mutation of the α2 subunit, on the other hand abolishes the effects of diazepam on measures of anxiety (Low et al.,
2000) and conditioned fear (Morris et al., 2006) in mice. Morris et al. (2006), however, observed that a compound with benzodiazepine agonist properties specific to α2, α3, and
α5 subunit-containing receptors still reduced conditioned suppression in these mice.
While point mutation of the α3 subunit does not abolish the anxiolytic properties of diazepam in mice (Low et al., 2000), a benzodiazepine inverse agonist active specifically at receptors containing α3 subunits is anxiogenic in rats (Atack et al., 2005).
An investigation of a benzodiazepine inverse agonist specific to α5 subunit-containing receptors, however, revealed no anxiogenic properties (Dawson et al., 2006).
Benzodiazepine inverse agonists with specificity for α5 subunit-containing receptors do, 185 Chapter 4. General Discussion
however, improve encoding and recall for the Morris water maze, a hippocampus- dependent spatial learning task, in rats, without producing the pro-convulsant effects typical of non-subunit-specific inverse agonists, and without altering motor coordination (Chambers et al., 2003; Collinson et al., 2006; Dawson et al., 2006).
Together, these findings suggest that benzodiazepine activity at receptors containing α1 subunit mediates sedative and convulsant effects, activity at receptors containing α2 and
α3 subunits mediates anxiety-related effects, and activity at receptors containing α5 subunits mediates mnemonic/amnesic effects of benzodiazepine ligands.
Because of the apparently distinct functions of benzodiazepine binding sites on different GABAA receptor subtypes, it would be of interest to investigate whether the effects of FG7142 on expression of overexpectation could be replicated with benzodiazepine inverse agonists specific to these different subtypes. There are specific reasons to suspect that receptors containing α2 subunits would be likely to mediate the effect of FG7142 on expression of overexpectation and extinction. First, Marowsky et al. (2004) report that inhibitory signalling in BLA is mediated largely by α1 and α2 subunit-containing receptors, while in CeA, this signalling is mediated almost entirely by receptors containing the α2 subunit. Second, Heldt and Ressler (2007) found that changes in GABAA receptor subunit expression in the CeA after extinction were only seen for the α2 subunit. Finally, Morris et al. (2006) found that the attenuation of a fear
CR by diazepam was mediated by receptors containing α2 subunits. If these receptors are involved in the attenuation of fear by a benzodiazepine agonist, they may also be involved in the attenuation of inhibition of fear by a benzodiazepine inverse agonist.
The hypotheses considered thus far in this section assume that FG7142 achieves its effect on the expression of fear extinction and overexpectation by modulating GABAA receptors that are directly responsible for inhibiting fear-related neural transmission. 186 Chapter 4. General Discussion
However, this effect may be mediated by indirect effects of FG7142 on other neurotransmitter systems. FG7142 enhances monoaminergic transmission in PFC.
FG7142 causes increased dopamine turnover in monkey PFC at a dose of 0.2 mg/kg
(Murphy et al., 1996) and in rat PFC at doses of 5-30 mg/kg (Tam & Roth, 1985;
Bassareo et al., 1996; Murphy et al., 1996). Evans et al. (2006) found that indices of mPFC serotonin metabolism in rats, including L-tryptophan, serotonin, and 5- hydroxyindoleacetic acid concentrations, were dose-dependently increased after injection of FG7142.
Sarter, Bruno, and Bernston (2001) propose that the effect of FG7142 on neurotransmitter systems in the cortex induces a “hyper-attentional” state. Specifically, they propose that FG7142-induced alterations in monoaminergic transmission dysregulate cholinergic transmission in the cortex, causing enhanced or biased processing of contextual signals related to fear and anxiety. This may explain why the effect of FG7142 on expression of extinction is context-specific (Harris & Westbrook,
1998) and also why general effects of FG7142 on fear expression were not observed in this thesis. When the meaning of a CS is made ambiguous by association with both a conditioning memory and an extinction memory, a bias towards processing fear-related aspects of a context may aid the selection of the former memory over the latter.
However, when a CS is unambiguously associated with a feared US (e.g. when it is simply conditioned and not subjected to any decremental training procedure), its presentation should elicit the conditioning memory regardless of co-present contextual information. Thus, modulation of contextual processing would not necessarily alter the
CR in such a situation. According to this theory, FG7142 would have no effect on expression of overexpectation in a sufficiently unique test context, provided that test context has no anxiogenic properties.
187 Chapter 4. General Discussion
Enhanced dopaminergic transmission in mPFC may also disinhibit fear responses by impairing PFC regulation of neural activity in amygdala. Rosenkranz and Grace (2001) found that systemic administration of the dopamine receptor agonist apomorphine attenuated mPFC – BLA interactions. For example, apomorphine prevented electrical stimulation of mPFC from exciting BLA inhibitory interneurons. While it is not clear whether this effect resulted from apomorphine action at dopamine receptors in mPFC or through interaction with additional brain structures, it is consistent with the general hypothesis that hyper-activation of dopamine receptors reduces regulation of mPFC – amygdala interactions. Furthermore, FG7142 itself disrupts vmPFC-amygdala interactions. Stevenson, Halliday, Marsden, and Mason (2007) recorded unit and field potential activity in the BA and vmPFC of anaesthetised rats after intravenous injection of vehicle and doses of 0.33 – 10 mg/kg FG7142. Doses ranging from 1 - 10 mg/kg disrupted the coherence of unit activity both within and between BA and vmPFC. Doses of 0.33 - 10 mg/kg reduced coherence of low frequency field potential activity between
BA and vmPFC. This low-frequency activity presumably reflected, at least in part, burst firing, which was also disrupted in both structures by doses of 1 -1 0 mg/kg.
The findings reported by Stevenson et al. (2007) suggest that FG7142 impairs burst activity-mediated coordination of vmPFC and BA neural activity. While there is no evidence that these effects are mediated by the effects of FG7142 on dopamine,
Rosenkranz and Grace’s (2001) findings suggest similar impairments by dopamine receptor activation. As FG7142 increases dopamine turnover in mPFC (Tam & Roth,
1985; Bassareo et al., 1996; Murphy et al., 1996), this may be a mechanism by which
FG7142 interferes with expression of fear extinction and overexpectation. This leads to testable hypotheses that the effects of FG7142 on expression of fear overexpectation and extinction may be blocked by dopamine receptor antagonists, and mimicked by
188 Chapter 4. General Discussion
dopamine receptor agonists, including following microinfusions of such drugs into the mPFC. Furthermore, the effects of FG7142 on dopamine turnover in mPFC in rats can be mimicked by injection of the benzodiazepine inverse agonist α3IA, which is selective for GABAA receptors containing the α3 subunit (Atack et al., 2005). Thus if the hypotheses mentioned above were supported by experimentation, it would also lead to the hypothesis that the effect of FG7142 on expression of fear extinction and overexpectation may be mediated specifically by GABAA receptors containing the α3 subunit.
6. Conclusion
This thesis represents the first investigation that the author is aware of examining the neurobiological mechanisms of expression of overexpectation. The main finding was that FG7142, a benzodiazepine partial inverse agonist, prevented expression of overexpectation of conditioned fear, as measured by freezing. FG7142 did not appear to have any effect on expression of the freezing CR after conditioning or after an associative blocking procedure. When considered alongside data indicating that FG7142 increases freezing to an extinguished CS, but does not increase freezing after latent inhibition (Harris & Westbrook, 1998; Kim et al., 2006), this suggests that enhancement of fear CRs by FG7142 is restricted to situations in which negative predictive error, as described by the Rescorla-Wagner model, leads to decrements in conditioned fear.
These findings add to evidence that overexpectation, like extinction, leads to new learning that partially masks fear during subsequent CS presentations. Furthermore, it is consistent with the hypothesis that, as in extinction, this mask on fear is neither permanent nor immutable. This suggests that this mask is a general result of negative prediction error, rather than the specific result of the pairing of a CS with the absence of
189 Chapter 4. General Discussion
a US. As some research (e.g. Delamater, 1996; Rescorla, 1999) suggests that specific
CS - US associations can be activated after extinction and overexpectation training it is suggested that negative prediction error results in the formation of an inhibitory CS- response association, rather than an inhibitory CS - US association. The CS continues to be able to activate a representation of the US, but also activates inhibition of the conditioned emotional response. FG7142 interferes with expression of this inhibitory
CS - response association.
The exact pharmacological mechanisms by which FG7142 prevents expression of extinction and overexpectation remain to be determined. Recent research suggests that
GABAergic transmission in several amygdala nuclei may contribute to the expression of extinction. Because FG7142 reduces the sensitivity of GABAA receptors to GABA, it may directly impede mechanisms of fear inhibition within the amygdala. Current neuroanatomical models also implicate projections from the vmPFC to the amygdala in expression of fear extinction. FG7142 appears to disrupt coherence of neural activity between the amygdala and vmPFC. This suggests that FG7142 may act to effectively
‘disconnect’ these two structures, thus providing another putative mechanism by which it impairs expression of extinction and overexpectation.
The effect of FG7142 may result from effects on other neurotransmitter systems that occur downstream of its modulation of GABAA receptors. In particular, FG7142 raises dopamine and serotonin levels in mPFC. This may disrupt mPFC - amygdala interactions responsible for expression of fear extinction and overexpectation. This may also modify attention in a manner that biases expression of fear. Further investigation of these possibilities will enhance understanding of the neurobiological mechanisms of learning, expression, and inhibition of fear. This would have possible clinical implications for the prevention and treatment of fear and anxiety disorders.
190
References
Akirav, I., Raizel, H., & Maroun, M. (2006). Enhancement of conditioned fear
extinction by infusion of the GABAA agonist muscimol into the rat prefrontal
cortex and amygdala. European Journal of Neuroscience, 23, 758-764.
Alvarez, R. P., Johnson, L., & Grillon, C. (2007). Contextual-specificity of short-delay
extinction in humans: Renewal of fear-potentiated startle in a virtual
environment. Learning & Memory, 14, 247-253.
Anglada-Figueroa, D. & Quirk, G. J. (2005). Lesions of the basal amygdala block
expression of conditioned fear but not extinction. The Journal of Neuroscience,
25, 9680-9685.
Atack, J. R., Hutson, P. H., Collinson, N., Marshall, G., Bentley, G., Moyes, C., Cook,
S. M., Collins, I., Wafford, K., McKernan, R. M., & Dawson, G. R. (2005).
Anxiogenic properties of an inverse agonist selective for α3 subunit-containing
GABAA receptors. British Journal of Pharmacology, 144, 357-366.
Aydin, A., & Pearce, J.M. (1997). Some determinants of response summation. Animal
Learning & Behavior, 25, 108-121.
Ayres, J. J. B. (1998). Fear conditioning and avoidance. In O’Donohue, W. T. (Ed.),
Learning and Behavior Therapy, 122-145. Needham Heights, Massachusetts:
Allyn & Bacon.
Bailey, D. J., Kim, J. J., Sun, W., Thompson, R. F., & Helmstetter, F. J. (1999).
Acquisition of fear conditioning in rats requires the synthesis of mRNA in the
amygdala. Behavioral Neuroscience, 113, 276-282.
Barad, M. (2006). Is extinction of fear erasure or inhibition? Why both, of course.
Learning & Memory, 13, 108-109.
191
Barad, M., Gean, P., & Lutz, B. (2006). The role of the amygdala in the extinction of
conditioned fear. Biological Psychiatry, 60, 322-328.
Bassareo, V., Tanda, G., Petromilli, P., Giua, C., & Di Chiara. G. (1996). Non-
psychostimulant drugs of abuse and anxiogenic drugs activate with differential
selectivity dopamine transmission in the nucleus accumbens and in the medial
prefrontal cortex of the rat. Psychopharmacology, 124, 293-299.
Bauer, E. P., Schafe, G. E., & LeDoux, J. E. (2002). NMDA receptors and L-type
voltage-gated calcium channels contribute to long-term potentiation and
different components of fear memory formation in the lateral amygdala. The
Journal of Neuroscience, 22, 5239-5249.
Beitz, A. J. (1982). The organization of afferent projections to the midbrain
periaqueductal gray of the rat. Neuroscience, 7, 133-159.
Berretta, S., Pantazopoulos, H., Caldera, M., Pantazopoulos, P., & Pare, D. (2005).
Infralimbic cortex activation increases c-Fos expression in intercalated neurons
of the amygdala. Neuroscience, 132, 943-953.
Berlau, D. J. & McGaugh, J. L. (2006). Enhancement of extinction memory
consolidation: the role of the noraderenergic and GABAergic systems within the
basolateral amygdala. Neurobiology of Learning and Memory, 86, 123-132.
Blaisdell, A. P., Denniston, J. C., & Miller, R. R. (1999). Post-training shifts in the
overshadowing stimulus-unconditioned stimulus interval alleviates the
overshadowing deficit. Journal of Experimental Psychology: Animal Behavior
Processes, 25, 18-27.
Blaisdell, A. P., Denniston, J. C., & Miller, R. R. (2001). Recovery from the
overexpectation effect: Contrasting performance-focused and acquisition-
192
focused models of retrospective revaluation. Animal Learning and Behavior, 29,
367-380.
Blaisdell, A. P., Gunther, L. M., & Miller, R. R. (1999). Recovery from blocking
achieved by extinguishing the blocking CS. Animal Learning and Behavior, 27,
63-76.
Bourtchuladze, R., Frenguelli, B., Blendy, J., Cioffi, D., Schutz, G., & Silvi, A. J.
(1994). Deficient long-term memory in mice with a targeted mutation of the
cAMP-responsive element-binding protein. Cell, 79, 59-68.
Bouton, M. E. (1984). Differential control by context in the inflation and reinstatement
paradigms. Journal of Experimental Psychology: Animal Behavior Processes,
10, 56-74.
Bouton, M. E. (1993). Context, time, and memory retrieval in the interference
paradigms of Pavlovian learning. Psychological Bulletin, 114, 80-99.
Bouton, M. E. (2004). Context and behavioral processes in extinction. Learning &
Memory, 11, 485-494.
Bouton, M. E. & Bolles, R. C. (1979). Contextual control of the extinction of
conditioned fear. Learning and Motivation, 10, 445-466.
Bouton, M. E. & Bolles, R. C. (1980). Conditioned fear assessed by freezing and by the
suppression of three different baselines. Animal Learning & Behavior, 8, 429-
434.
Bouton, M. E., Kenney, F. A., & Rosengard, C. (1990). State-dependent fear extinction
with two benzodiazepine tranquilizers. Behavioral Neuroscience, 104, 44-55.
Bouton, M. E. & Ricker, S. T. (1994). Renewal of conditioned responding in a second
context. Animal Learning & Behavior, 22, 317-324.
193
Brown, J. S., Kalish, H. I., & Farber, I. E. (1951). Conditioned fear as revealed by
magnitude of startle response to an auditory stimulus. Journal of Experimental
Psychology, 41, 317-328.
Burgos-Robles, A., Vidal-Gonzalez, I., Santini, E., & Quirk, G. J. (2007). Consolidation
of fear extinction requires NMDA receptor-dependent bursting in the
ventromedial prefrontal cortex. Neuron, 53, 871-880.
Cahill, L. & McGaugh, J. L. (1998). Mechanisms of emotional arousal and lasting
declarative memory. Trends in Neurosciences, 21, 294-299.
Campeau, S., Miserendino, M. J. D., & Davis, M. (1992). Intra-amygdala infusion of
the N-methyl-D-aspartate receptor antagonist AP5 blocks acquisition but not
expression of fear-potentiated startle to an auditory conditioned stimulus.
Behavioral Neuroscience, 106, 569-574.
Chambers, M. S., Atack, J. R., Broughton, H. B., Collinson, N., Cook, S., Dawson, G.
R., Hobbs, S. C., Marshall, G., Maubach, K. A., Pillai, G. V., Reeve, A. J., &
MacLeod, A. M. (2003). Identification of a novel, selective GABAA α5 receptor
inverse agonist which enhances cognition. Journal of Medicinal Chemistry, 46,
2227-2240.
Chhatwal, J. P., Davis, M., Maguschak, K. A., & Ressler, K. J. (2005). Enhancing
cannabinoid neurotransmission augments the extinction of conditioned fear.
Neuropsychopharmacology, 30, 516-524.
Chhatwal, J. P., Myers, K. M., Ressler, K. J., & Davis, M. (2005). Regulation of
gephyrin and GABAA receptor binding within the amygdala after fear
acquisition and extinction. The Journal of Neuroscience, 25, 502-506.
194
Chiba, T., Kayahara, T. & Nakano, K. (2001). Efferent projections of infralimbic and
prelimbic areas of the medial prefrontal cortex in the Japanese monkey, Macaca
fuscata. Brain Research, 888, 83-101.
Cicala, G. A., Azorlosa, J. L., Estall, L. B., & Grant, S. J. (1990). Endogenous opioids
interfere with Pavlovian second-order conditioning. Psychological Science, 1,
312-315.
Cole, B. J., Hillmann, M., Seidelmann, D., Klewer, M., & Jones, G. H. (1995). Effects
of benzodiazepine receptor partial inverse agonists in the elevated plus maze test
of anxiety in the rat. Psychopharmacology, 121, 118-126.
Cole, S. & McNally, G. P. (2007). Opioid receptors mediate direct predictive fear
learning: Evidence from one-trial blocking. Learning & Memory, 14, 229-235.
Collins, D. J. & Shanks, D. R. (2006). Summation in causal learning: Elemental
processing or configural generalization? The Quarterly Journal of Experimental
Psychology, 59, 1524-1534.
Collins, D. R. & Pare, D. (2000). Differential fear conditioning induces reciprocal
changes in the sensory responses of lateral amygdala neurons to the CS+ and CS-
. Learning & Memory, 7, 97-103.
Collinson, N., Atack, J. R., Laughton, P., Dawson, G. R., & Stephens, D. N. (2006). An
inverse agonist selective for α5 subunit-containing GABAA receptors improves
encoding and recall but not consolidation in the Morris water maze.
Psychopharmacology, 188, 619-628.
Corcoran, K. A., Desmond, T. J., Frey, K. A., & Maren, S. (2005). Hippocampal
inactivation disrupts the acquisition and contextual encoding of fear extinction.
The Journal of Neuroscience, 25, 8978-8987.
195
Corcoran, K. A. & Maren, S. (2001). Hippocampal inactivation disrupts contextual
retrieval of fear memory after extinction. The Journal of Neuroscience, 21,
1720-1726.
Corcoran, K. A. & Maren, S. (2004). Factors regulating the effects of hippocampal
inactivation on renewal of conditional fear after extinction. Learning & Memory,
11, 598-603.
Corcoran, K. A. & Quirk, G. J. (2007). Activity in prelimbic cortex is necessary for the
expression of learned, but not innate, fears. The Journal of Neuroscience, 27,
840-844.
Corda, M. G., Blaker, W. D., Mendelson, W. B., Guidotti, A., & Costa, E. (1983). Beta-
carbolines enhance shock-induced suppression of drinking in rats. Proceedings
of the National Academy of Sciences of the United States of America, 80, 2072-
2076.
Cumming, G., & Finch, S. (2005). Inference by eye: Confidence intervals, and how to
read pictures of data. American Psychologist, 60, 170–180.
Davis, M., Falls, W. A., Campeau, S., & Kim M. (1993). Fear-potentiated strartle: A
neural and pharmacological analysis. Behavioural Brain Research, 58, 175-198.
Davis, M. & Whalen, P. J. (2001). The amygdala: Vigilance and emotion. Molecular
Psychiatry, 6, 13-34.
Dawson, G. R., Maubach, K. A., Collinson, N., Cobain, M., Everitt, B. J., Macleod, A.
M., Choudhury, H. I., McDonald, L. M., Pillai, G., Rycroft, W., Smith, A. J.,
Stermfeld, F., Tattersall, F. D., Wafford, K. A., Reynolds, D. S., Seabrook, G.
R., & Atack, J. R. (2006). An inverse agonist selective for α5 subunit-containing
GABAA receptors enhances cognition. The Journal of Pharmacology and
Experimental Therapeutics, 316, 1335-1345.
196
Delamater, A. R. (1996). Effects of several extinction treatments upon the integrity of
Pavlovian stimulus outcome associations. Animal Learning & Behavior, 24,
437-449.
Delamater, A. R. (2004). Experimental extinction in Pavlovian conditioning:
Behavioural and neuroscience perspectives. Quarterly Journal of Experimental
Psychology Section B, 57, 97-132.
Delamater, A. R., Campese, V., & Westbrook, R. F. (In Press). Renewal and
spontaneous recovery, but not latent inhibition, are mediated by GABA in
appetitive conditioning. Journal of Experimental Psychology: Animal Behavior
Processes.
Delamater, A. R., Sosa, W., & Katz, M. (1999). Elemental and configural processes in
patterning discrimination learning. The Quarterly Journal of Experimental
Psychology, 52B, 97-124.
Denniston, J. C., Chang, R. C., & Miller, R. R. (2003). Massive extinction treatment
attenuates the renewal effect. Learning and Motivation, 34, 68-86.
Dickinson, A. & Burke, J. (1996). Within-compound associations mediate the
retrospective revaluation of causality judgements. The Quarterly Journal of
Experimental Psychology, 49B, 60-80.
Dickinson, A. & Pearce, J. M. (1977). Inhibitory interactions between appetitive and
aversive stimuli. Psychological Bulletin, 84, 690-711.
Dorow, R., Horowski, R., Paschelke, G., & Amin, M. (1983). Severe anxiety induced
by FG 7142, a beta-carboline ligand for benzodiazepine receptors. Lancet, 322,
98-99.
Estes, W. K. & Skinner, B. F. (1941). Some quantitative properties of anxiety. Journal
of Experimental Psychology, 29, 390-400.
197
Evans, A. K., Abrams, J. K., Bouwknecht, J. A., Knight, D. M., Shekhar, A., & Lowry,
C. A. (2006). The anxiogenic drug FG-7142 increases serotonin metabolism in
the rat medial prefrontal cortex. Pharmacology Biochemistry and Behavior, 84,
266-274.
Evans, A. K. & Lowry, C. A. (2007). Pharmacology of the β-carboline FG-7142, a
partial inverse agonist at the benzodiazepine allosteric site of the GABAA
receptor: Neurochemical, neurophysiological, and behavioral effects. CNS Drug
Reviews, 13, 475-501.
Falls, W. A., Miserendino, M. J. D., & Davis, M. (1992). Extinction of fear-potentiated
startle: Blockade by infusion of an NMDA antagonist into the amygdala. The
Journal of Neuroscience, 12, 854-863.
Fanselow, M. S. (1980). Conditional and unconditional components of post-footshock
freezing in rats. Pavlovian Journal of Biological Science, 15, 177-182.
Fanselow, M. S. (1981). Naloxone and Pavlovian fear conditioning. Learning and
Motivation, 12, 398-419.
Fanselow, M. S. (1998). Pavlovian conditioning, negative feedback, and blocking:
Mechanisms that regulate association formation. Neuron, 20, 625-627.
Feenstra, M. G. P., Vogel, M., Botterblom, M. H. A., Joosten, R. N. J. M. A., & de
Bruin, J. P. C. (2001). Dopamine and noradrenaline efflux in the rat prefrontal
cortex after classical aversive conditioning to an auditory cue. European Journal
of Neuroscience, 13, 1051-1054.
Fritschy, J. M., Schweizer, C., Brunig, I., & Luscher, B. (2003). Pre- and post-synaptic
mechanisms regulating the clustering of type A γ-aminobutyric acid receptors
(GABAA receptors). Biochemical Society Transactions, 31, 889-892.
198
Frank, D. A. & Greenberg, M. E. (1994). CREB: A mediator of long-term memory from
mollusks to mammals. Cell, 79, 5-8.
Garakani, A., Mathew, S. J., & Charney, D. S. (2006). Neurobiology of anxiety
disorders and implications for treatment. The Mount Sinai Journal of Medicine,
73, 941-949.
Garcia, R., Chang, C., & Maren, S. (2006). Electrolytic lesions of the medial prefrontal
cortex do not interfere with long-term memory of extinction of conditioned fear.
Learning & Memory, 13, 14-17.
Garcia, R., Vouimba, R., Baudry, M., & Thompson, R. F. (1999). The amygdala
modulates prefrontal cortex activity relative to conditioned fear. Nature, 402,
294-296.
Gewirtz, J. C., Falls, W. A., & Davis, M. (1997). Normal conditioned inhibition and
extinction of freezing and fear potentiated startle following electrolytic lesions
of medial prefrontal cortex in rats. Behavioral Neuroscience, 111, 712-726.
Goosens, K. A. & Maren, S. (2003). Pretraining NMDA receptor blockade in the
basolateral complex, but not the central nucleus, of the amygdala prevents
savings of conditional fear. Behavioral Neuroscience, 117, 738-750.
Harris, J. A., Jones, M. L., Bailey, G. K., & Westbrook, R. F. (2000). Contextual control
over conditioned responding in an extinction paradigm. Journal of Experimental
Psychology: Animal Behavior Processes, 26, 174-185.
Harris, J. A. & Westbrook, R. F. (1998). Evidence that GABA transmission mediates
context-specific extinction of learned fear. Psychopharmacology, 140, 105-115.
Harris, J. A. & Westbrook, R. F. (1999). The benzodiazepine midazolam does not
impair Pavlovian fear conditioning but regulates when and where fear is
199
expressed. Journal of Experimental Psychology: Animal Behavior Processes,
25, 236-246.
Hays, W. L. (1972). Statistics for the social sciences. Holt, Rinehart and Winston, New
York.
Heldt, S. A. & Ressler, K. J. (2007) Training-induced changes in the expression of
GABAA-associated genes in the amygdala after the acquisition and extinction of
Pavlovian fear. European Journal of Neuroscience, 26, 3631-3644.
Helmstetter, F. J. & Bellgowan, P. S. (1994). Effects of muscimol applied to the
basolateral amygdala on acquisition and expression of contextual fear
conditioning in rats. Behavioral Neuroscience, 108, 1005-1009.
Herry, C., Trifillief, P., Micheau, J., Luthi, A., & Mons, N. (2006). Extinction of
auditory fear conditioning required MAPK/ERK activation in the basolateral
amygdala. European Journal of Neuroscience, 24, 261-269.
Hitchcock, J. M. & Davis, M. (1986). Lesions of the amygdala, but not of the
cerebellum or red nucleus, block conditioned fear as measured with the
potentiated startle paradigm. Behavioral Neuroscience, 100, 11-22.
Hitchcock, J. M. & Davis, M. (1987). Fear-potentiated startle using an auditory
conditioned stimulus: Effects of lesions of the amygdala. Physiology and
Behavior, 39, 403-408.
Hobin, J. A., Goosens, K. A., & Maren, S. (2003). Context-dependent neuronal activity
in the lateral amygdala represents fear memories after extinction. The Journal of
Neuroscience, 23, 8410-8416.
Hosoba, T., Iwanaga, M. & Seiwa, H. (2001). The effect of UCS inflation and deflation
procedures on ‘fear’ conditioning. Behaviour Research and Therapy, 39, 465-
475.
200
Hugues, S., Chessel, A., Lena, I., Marsault, R., & Garcia, R. (2006). Prefrontal infusion
of PD098059 immediately after fear extinction training blocks extinction-
associated prefrontal synaptic plasticity and decreases prefrontal ERK2
phosphorylation. Synapse, 60, 280-287.
Hugues, S., Deschaux, O., & Garcia, R. (2004). Postextinction infusion of a mitogen-
activated protein kinase inhibitor into the medial prefrontal cortex impairs
memory of extinction of conditioned fear. Learning & Memory, 11, 540-543.
Impey, S., Obrieton, K., Wong, S. T., Poser, S., Yano, S., Wayman, G., Deloulme, J. C.,
Chan, G., & Storm, D. R. (1998). Cross talk between ERK and PKA is required
for Ca2+ stimulation of CREB-dependent transcription and ERK nuclear
translation. Neuron, 21, 869-883.
Ishikawa, A. & Nakamura, S. (2003). Convergence and interaction of hippocampal and
amygdalar projections within the prefrontal cortex in the rat. The Journal of
Neuroscience, 23, 9987-9995.
Iwata, J., LeDoux, J. E., Meeley, M. P., Arneric, S., & Reis, D. J. (1986). Intrinsic
neurons in the amygdaloid field projected to by the medial geniculate body
mediate emotional responses conditioned to acoustic stimuli. Brain Research,
383, 195-214.
Kamin, L. J. (1968). “Attention-like” processes in classical conditioning. In Jones, M.
R. (Ed.), Miami Symposium on the Prediction of Behavior: Aversive Stimulation,
9-31. Miami: University of Miami Press.
Kamprath, K., Marsicano, G., Tang, J., Monory, K., Bisogno, T., Di Marzo, V., Lutz,
B., & Wotjak, C. T. (2006). Cannabinoid CB1 receptor mediates fear extinction
via habituation-like processes. The Journal of Neuroscience, 26, 6677-6686.
201
Kehoe, E. J. (1988). A layered network model of associative learning: Learning to learn
and configuration. Psychological Review, 95, 411-433.
Kehoe, E. J. & White, N. E. (2004). Overexpectation: Response loss during sustained
stimulus compounding in the rabbit nictitating membrane preparation. Learning
& Memory, 11, 476-483.
Kellicutt, M. H. & Schwartzbaum, J. S. (1963). Formation of a conditioned emotional
response (CER) following lesions of the amygdaloid complex in rats.
Psychological Reports, 12, 351-358.
Khallad, Y. & Moore, J. (1996). Blocking, unblocking, and overexpectation in
autoshaping with pigeons. Journal of the Experimental Analysis of Behavior, 65,
575-591.
Kim, J. H., McNally, G. P., & Richardson, R. (2006). Recovery of fear memories in
rats: Role of gamma-amino butyric acid (GABA) in infantile amnesia.
Behavioral Neuroscience, 120, 40-48.
Kim, J. H. & Richardson, R. (2007). A developmental dissociation of context and
GABA effects on extinguished fear in rats. Behavioral Neuroscience, 121, 131-
139.
Kim, M., Campeau, S., Falls, W. A., & Davis, M. (1993). Infusion of the non-NMDA
receptor antagonist CNQX into the amygdala blocks the expression of fear-
potentiated startle. Behavioral and Neural Biology, 59, 5-8.
Konorski, J. (1967). Integrative Activity of the Brain. Chicago: University of Chicago
Press.
Kremer, E. F. (1978). The Rescorla-Wagner model: Losses in associative strength in
compound conditioned stimuli. Journal of Experimental Psychology: Animal
Behavior Processes, 4, 22-36.
202
LaBar, K. S., Gatenby, J. C., Gore, J. C., LeDoux, J. E., & Phelps, E. A. (1998). Human
amygdala activation during conditioned fear acquisition and extinction: A
mixed-trial fMRI study. Neuron, 20, 937-945.
LaBar, K. S. & Phelps, E. A. (2005). Reinstatement of conditioned fear in humans is
context-dependent and impaired in amnesia. Behavioral Neuroscience, 119, 677-
686.
Lattal, K. M. & Nakajima, S. (1998). Overexpectation in appetitive Pavlovian and
instrumental conditioning. Animal Learning & Behavior, 26, 351-360.
Laviolette, S. R. & Grace, A. A. (2006). Cannabinoids potentiate emotional learning
plasticity in neurons of the medial prefrontal cortex through basolateral
amygdala inputs. The Journal of Neuroscience, 26, 6458-6468.
Laviolette, S. R., Lipski, W. J., & Grace, A. A. (2005). A subpopulation of neurons in
the medial prefrontal cortex encodes emotional learning with burst and
frequency codes through a dopamine D4 receptor-dependent basolateral
amygdala input. The Journal of Neuroscience, 25, 6066-6075.
Leaf, R. C. & Leaf, S. R. P. (1966). Recovery time as a measure of degree of
conditioned suppression. Psychological Reports, 18, 265-266.
Leaf, R. C. & Muller, S. A. (1965). Simple method for CER conditioning and
measurement. Psychological Reports, 17, 211-215.
Lebron, K., Milad, M. R., & Quirk, G. J. (2004). Delayed recall of fear extinction in rats
with lesions of ventral medial prefrontal cortex. Learning & Memory, 11, 544-
548.
LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience, 23,
155-184.
203
LeDoux, J.E., Farb, C., & Ruggiero, D. A. (1990). Topographic organization of neurons
in the acoustic thalamus that project to the amygdala. The Journal of
Neuroscience, 10, 1043-1054.
LeDoux, J. E., Iwata, J., Cicchetti, P., & Reis, D. J. (1988). Different projections of the
central amygdaloid nucleus mediate autonomic and behavioral correlates of
conditioned fear. The Journal of Neuroscience, 8, 2517-2529.
Lee, H. & Kim, J. J. (1998). Amygdalar NMDA receptors are critical for new fear
learning in previously fear-conditioned rats. The Journal of Neuroscience, 18,
8444-8454.
Lin, C., Yeh, S., Leu, T., Chang, W., Wang, S, & Gean, P. (2003). Identification of
calcineurin as a key signal in the extinction of fear memory. The Journal of
Neuroscience, 23, 1574-1579.
Lin, C., Yeh, S., Lin, C., Lu, K., Leu, T., Chang, W., & Gean, P. (2001). A role for the
PI-3 kinase signalling pathway in fear conditioning and synaptic plasticity in the
amygdala. Neuron, 31, 841-851.
Lin, C., Yeh, S., Lu, H., & Gean, P. (2003). The similarities and diversities of signal
pathways leading to consolidation of conditioning and consolidation of
extinction of fear memory. The Journal of Neuroscience, 23, 8310-8317.
Lovibond, P. F., Davis, N. R., & O’Flaherty, A. S. (2000). Protection from extinction in
human fear conditioning. Behaviour Research and Therapy, 38, 967-983.
Low, K., Crestani, F., Keist, R., Benke, D., Brunig, I., Benson, J. A., Fritschy, J.,
Rulicke, T., Bluethmann, H., Mohler, H., & Rudolph, U. (2000). Molecular and
neuronal substrate for the selective attenuation of anxiety. Science, 290, 131-
134.
204
Lu, K., Walker, D. L., & Davis, M. (2001). Mitogen-activated protein kinase cascade in
the basolateral nucleus of amygdala is involved in extinction of fear-potentiated
startle. The Journal of Neuroscience, 21: RC162, 1-5.
Mackintosh, N. J. (1975a). A theory of attention: Variations in the associability of
stimuli with reinforcement. Psychological Review, 82, 276-298.
Mackintosh, N. J. (1975b). Blocking of conditioned suppression: Role of the first
compound trial. Journal of Experimental Psychology: Animal Behavior
Processes, 1, 335-345.
Mao, S., Hsiao, Y., & Gean, P. (2006). Extinction training in conjunction with a partial
agonist of the glycine site on the NMDA receptor erases memory trace. The
Journal of Neuroscience, 26, 8892-8899.
Maren, S. (2000). Auditory fear conditioning increases CS-elicited spike firing in lateral
amygdala neurons even after extensive overtraining. European Journal of
Neuroscience, 12, 4047-4054.
Maren, S., Aharonov, G., Stote, D. L., & Fanselow, M. S. (1996). N-methyl-D-aspartate
receptors in the basolateral amygdala are required for both acquisition and
expression of conditional fear in rats. Behavioral Neuroscience, 110, 1365-1374.
Maren, S. & Chang, C. (2006). Recent fear is resistant to extinction. Proceedings of the
National Academy of Sciences of the United States of America, 103, 18020-
18025.
Maren, S. & Quirk, G. J. (2004). Neuronal signalling of fear memory. Nature Reviews
Neuroscience, 5, 844-852.
Marowsky, A., Fritschy, J., & Vogt, K. E. (2004). Functional mapping of GABAA
receptor subtypes in the amygdala. European Journal of Neuroscience, 20,
1281-1289.
205
Marsicano, G., Wotjak, C. T., Azad, S. C., Bisogno, T., Rammes, G., Cascio, M. G.,
Hermann, H., Tang, J., Hofmann, C., Zieglgansberger, W., Di Marzo, V., &
Lutz, B. (2002). The endogenous cannabinoid system controls extinction of
aversive memories. Nature, 418, 530-534.
Matzel, L. D., Schachtman, T. R., & Miller, R. R. (1985). Recovery of an
overshadowed association achieved by extinction of the overshadowing
stimulus. Learning & Motivation, 16, 398-412.
McDonald, A. J. (1987). Organization of amygdaloid projections to the mediodorsal
thalamus and prefrontal cortex: A fluorescence retrograde transport study in the
rat. The Journal of Comparative Neurology, 262, 46-58.
McDonald, A. J., Mascagni, F., & Guo, L. (1996). Projections of the medial and lateral
prefrontal cortices to the amygdala: A phaseolus vulgaris leucoagglutinin study
in the rat. Neuroscience, 71, 55-75.
McKernan, R. M., Rosahl, T. W., Reynolds, D. S., Sur, C., Wafford, K. A., Atack, J. R.,
Farrar, S., Myers, J., Cook, G., Ferris, P., Garrett, L., Bristow, L., Marshall, G.,
Macauley, A., Brown, N., Howell, O., Moore, K. W., Carling, R. W., Street, L.
J., Castro J. L., Ragan, C. I., Dawson, G. R., & Whiting P. J. (2000). Sedative
but not anxiolytic properties of benzodiazepines are mediated by the GABAA
receptor α1 subtype. Nature Neuroscience, 3, 587-592.
McNally, G. P. & Cole, S. (2006). Opioid receptors in the midbrain periaqueductal gray
regulate prediction error during Pavlovian fear conditioning. Behavioral
Neuroscience, 120, 313-323.
McNally, G. P., Pigg, M., & Weidemann, G. (2004a). Blocking, unblocking, and
overexpectation of fear: A role for opioid receptors in the regulation of
Pavlovian association formation. Behavioral Neuroscience, 118, 111-120.
206
McNally, G. P., Pigg, M., & Weidemann, G. (2004b). Opioid receptors in the midbrain
periaqueductal gray regulate extinction of Pavlovian fear conditioning. Journal
of Neuroscience, 24, 6912-6919.
McNally, G. P. & Westbrook, R. F. (2003). Opioid receptors regulate the extinction of
Pavlovian fear conditioning. Behavioral Neuroscience, 117, 1292-1301.
McSweeney, F. K. & Swindell, S. (2002). Common processes may contribute to
extinction and habituation. The Journal of General Psychology, 129, 364-400.
Milad, M. R., & Quirk, G. J. (2002). Neurons in medial prefrontal cortex signal memory
for fear extinction. Nature, 420, 70-74.
Milad, M. R., Vidal-Gonzalez, I, & Quirk, G. J. (2004). Electrical stimulation of medial
prefrontal cortex reduces conditioned fear in a temporally specific manner.
Behavioral Neuroscience, 118, 389-395.
Milad, M. R., Wright, C. I., Orr, S. P., Pitman, R. K., Quirk, G. J., & Rauch, S. L.
(2007). Recall of fear extinction in humans activates the ventromedial prefrontal
cortex and hippocampus in concert. Biological Psychiatry, 62, 446-454.
Mineka, S., Mystkowski, J. L., Hladek, D., & Rodriguez, B. I. (1999). The effects of
changing contexts on return of fear following exposure therapy for spider fear.
Journal of Consulting and Clinical Psychology, 67, 599-604.
Miserendino, M. J. D., Sananes, C. B., Melia, K. R., & Davis, M. (1990). Blocking of
acquisition but not expression of conditioned fear-potentiated startle by NMDA
antagonists in the amygdala. Nature, 345, 716-718.
Mohler, H., Fritschy, J. M., & Rudolph, U. (2002). A new benzodiazepine
pharmacology. The Journal of Pharmacology and Experimental Therapeutics,
300, 2-8.
207
Mordkoff, A. M., Edelberg, R., & Ustick, M. (1967). The differential conditionability of
two components of the skin conductance response. Psychophysiology, 4, 40-47.
Morgan, M. A., Romanski, L. M., & LeDoux, J. E. (1993). Extinction of emotional
learning: Contribution of medial prefrontal cortex. Neuroscience Letters, 163,
109-113.
Morris, H. V., Dayson, G. R., Reynolds, D. S., Atack, J. R., & Stephens, D. N. (2006)
Both α2 and α3 GABAA receptor subtypes mediate the anxiolytic properties of
benzodiazepine site ligands in the conditioned emotional response paradigm.
European Journal of Neuroscience, 23, 2495-2504.
Moses, S. N., Houck, J. M., Martin, T., Hanlon, F. M., Ryan, J. D., Thoma, R. J.,
Weisend, M. P., Jackson, E. M., Pekkonen, E., & Tesche, C. D. (2007).
Dynamic neural activity recorded from human amygdala during fear
conditioning using magnetoencephalography. Brain Research Bulletin, 71, 452-
460.
Muller, J., Corodimas, K. P., Fridel, Z., & LeDoux, J. E. (1997). Functional inactivation
of the lateral and basal nuclei of the amygdala by muscimol infusion prevents
fear conditioning to an explicit conditioned stimulus and to contextual stimuli.
Behavioral Neuroscience, 111, 683-691.
Murphy, B. L., Arnsten, A. F., Jentsch, J. D., & Roth, R. H. (1996). Dopamine and
spatial working memory in rats and monkeys: Pharmacological reversal of
stress-induced impairment. The Journal of Neuroscience, 16, 7768-7775.
Myers, K. M. & Davis, M. (2007). Mechanisms of fear extinction. Molecular
Psychiatry, 12, 120-150.
208
Myers, K. M., Ressler, K. J., & Davis, M. (2006). Different mechanisms of fear
extinction dependent on length of time since fear extinction. Learning &
Memory, 13, 216-223.
Mystkowski, J. L., Craske, M. G., & Echiverri, A. M. (2002). Treatment context and
return of fear in spider phobia. Behavior Therapy, 33, 399-416.
Nader, K., Majidishad, P., Amaropanth, P., & LeDoux, J. E. (2001). Damage to the
lateral and central, but not other, amygdaloid nuclei prevents the acquisition of
auditory fear conditioning. Learning & Memory, 8, 156-163.
Niehoff, D. L., & Kumar, M. J. (1983). Benzodiazepine receptors: Localization in rat
amygdala. The Journal of Neuroscience, 3, 2091-2097.
Pape, H. & Stork, O. (2003). Genes and mechanisms in the amygdala involved in the
formation of fear memory. Annals of the New York Academy of Sciences, 985,
92-105.
Pare, D. & Collins, D. R. (2000). Neuronal correlates of fear in the lateral amygdala:
Multiple extracellular recordings in conscious cats. The Journal of
Neuroscience, 20, 2701-2710.
Pare, D., Quirk, G. J., & LeDoux, J. E. (2004). New vistas on amygdala networks in
conditioned fear. Journal of Neurophysiology, 92, 1-9.
Pearce, J. M. (2002). Evaluation and development of a connectionist theory of
configural learning. Animal Learning & Behavior, 30, 73-95.
Pearce, J. M. & Hall, G. (1980). A model for Pavlovian learning: Variations in the
effectiveness of conditioned but not unconditioned stimuli. Psychological
Review, 87, 532-552.
209
Pellow, S. & File, S. E. (1986). Anxiolytic and anxiogenic drug effects on exploratory
activity in an elevated plus maze: A novel test of anxiety in the rat.
Pharmacology Biochemistry and Behavior, 24, 525-529.
Pezze, M. A., Bast, T., & Feldon, J. (2003). Significance of dopamine transmission in
the rat medial prefrontal cortex for conditioned fear. Cerebral Cortex, 13, 371-
380.
Phelps, E. A., Delgado, M. R., Nearing, K. I., & LeDoux, J. E. (2004). Extinction
learning in humans: Role of the amygdala and vmPFC. Neuron, 43, 897-905.
Quirk, G. J. (2002). Memory for extinction of conditioned fear is long-lasting and
persists following spontaneous recovery. Learning & Memory, 9, 402-407.
Quirk, G. J., Likhtik, E., Pelletier, J. G., & Pare, D. (2003). Stimulation of medial
prefrontal cortex decreases the responsiveness of central amygdala output
neurons. The Journal of Neuroscience, 23, 8800-8807.
Quirk, G. J., Repa, C., & LeDoux, J. E. (1995). Fear conditioning enhances short-
latency auditory responses of lateral amygdala neurons: Parallel recordings in
the freely behaving rat. Neuron, 15, 1029-1039.
Quirk, G. J., Russo, G. K., Barron, J. L., & Lebron, K. (2000). The role of ventromedial
prefrontal cortex in the recovery of extinguished fear. The Journal of
Neuroscience, 20, 6225-6231.
Reijmers, L. G., Perkins, B. L., Matsuo, N., & Mayford, M. (2007). Localization of a
stable neural correlate of associative memory. Science, 317, 1230-1233.
Repa, J. C., Muller, J., Apergis, J., Desrochers, T. M., Zhou, Y., & LeDoux, J. E.
(2001). Two different lateral amygdala cell populations contribute to the
initiation and storage of memory. Nature Neuroscience, 4, 724-731.
210
Rescorla, R. A. (1970). Reduction in the effectiveness of reinforcement after prior
excitatory conditioning. Learning and Motivation, 1, 372-381.
Rescorla, R. A. (1973). Effect of US habituation following conditioning. Journal of
Comparative and Physiological Psychology, 82, 137-143.
Rescorla, R. A. (1982). Some consequences of associations between the excitor and the
inhibitor in a conditioned inhibition paradigm. Journal of Experimental
Psychology: Animal Behavior Processes, 8, 288-298.
Rescorla, R. A. (1993). Inhibitory associations between S and R in extinction. Animal
Learning & Behavior, 21, 327-336.
Rescorla, R.A. (1997). Summation: Assessment of a configural theory. Animal Learning
& Behavior, 25, 200-209.
Rescorla, R. A. (1999). Summation and overexpectation with qualitatively different
outcomes. Animal Learning & Behavior, 27, 50-62.
Rescorla, R. A. (2000). Extinction can be enhanced by a concurrent excitor. Journal of
Experimental Psychology: Animal Behavior Processes, 26, 251-260.
Rescorla, R. A. (2003). Protection from extinction. Learning & Behavior, 31, 124-132.
Rescorla, R. A. (2004). Spontaneous recovery. Learning & Memory, 11, 501-109.
Rescorla, R. A (2006). Spontaneous recovery from overexpectation. Learning &
Behavior, 34, 13-20.
Rescorla, R. A. (2007). Renewal after overexpectation. Learning & Behavior, 35, 19-
26.
Rescorla, R. A. & Heth, C. D. (1975). Reinstatement of fear to an extinguished
conditioned stimulus. Journal of Experimental Psychology: Animal Behavior
Processes, 104, 88-96.
211
Rescorla, R. A. & Wagner, A. R. (1972). A theory of Pavlovian conditioning:
Variations in the effectiveness of reinforcement and nonreinforcement. In Black,
A. H. & Prokasy, W. K. (Eds.), Classical Conditioning II: Current Research and
Theory, 64-99. New York: Appleton-Century-Crofts.
Ressler, K. J., Paschall, G., Zhou, X., & Davis, M. (2002). Regulation of synaptic
plasticity genes during consolidation of fear conditioning. The Journal of
Neuroscience, 22, 7892-7902.
Rizvi, T. A., Ennis, M., Behbehani, M. M., & Shipley, M. T. (1991). Connections
between the central nucleus of the amygdala and the midbrain periaqueductal
gray: Topography and reciprocity. The Journal of Comparative Neurology, 303,
121-131.
Rodgers, R. J., Cole, J. C., Aboualfa, K., & Stephenson, L. H. (1995).
Ethopharmacological analysis of the effects of putative ‘anxiogenic’ agents in
the mouse elevated plus-maze. Pharmacology Biochemistry and Behavior, 52,
805-813.
Rodrigues, S. M., Schafe, G. E., LeDoux, J. E. (2001). Intra-amygdala blockade of the
NR2B subunit of the NMDA receptor disrupts the acquisition but not the
expression of fear conditioning. The Journal of Neuroscience, 21, 6889-6896.
Rogan, M. T., Leon, K. S., Perez, D. L., & Kandel, E. R. (2005). Distinct neural
signatures for safety and danger in the amygdala and striatum of the mouse.
Neuron, 46, 309-320.
Rogan, M. T., Staubli, U. V., & LeDoux, J. E. (1997). Fear conditioning induces
associative long-term potentiation in the amygdala. Nature, 390, 604-607.
212
Romanski, L. M., Clugnet, M., Bordi, F., & LeDoux, J. E. (1993). Somatosensory and
auditory convergence in the lateral nucleus of the amygdala. Behavioral
Neuroscience, 107, 444-450.
Room, P. R., Russchen, F. T., Groenewegen, H. J., & Lohman, A. H. M. (1985).
Efferent connections of the prelimbic (area 32) and infralimbic (area 25)
cortices: An anterograde tracing study in the cat. Journal of Comparative
Neurology, 242, 40-55.
Rosen, L. B., Ginty, D. D., Weber, M. J., & Greenberg, M. E. (1994). Membrane
depolarization and calcium influx stimulate MEK and MAP kinase via
activation of ras. Neuron, 12, 1207-1221.
Rosenkranz, J. A. & Grace, A. A. (2001). Dopamine attenuates prefrontal cortical
suppression of sensory inputs to the basolateral amygdala of rats. The Journal of
Neuroscience, 21, 4090-4103.
Rosenkranz, J. A. & Grace, A. A. (2002). Cellular mechanisms of infralimbic and
prelimbic prefrontal cortical inhibition and dopaminergic modulation of
basolateral amygdala neurons in vivo. The Journal of Neuroscience, 22, 324-
337.
Rosenkranz, J. A., Moore, H., & Grace, A. A. (2003). The prefrontal cortex regulates
lateral amygdala neuronal plasticity and responses to previously conditioned
stimuli. The Journal of Neuroscience, 23, 11054-11064.
Rudolph, U., Crestani, F., Benke, D., Brunig, I., Benson, J. A., Fritschy, J. M., Martin,
J. R., Bluethmann, H., & Mohlner, H. (1999). Benzodiazepine actions mediated
by specific gamma-aminobutyric acid(A) receptor subtypes. Nature, 401, 796-
800.
213
Rumpel, S., LeDoux, J., Zador, A., & Malinow, R. (2005). Postsynaptic receptor
trafficking underlying a form of associative learning. Science, 308, 83-88.
Sacchetti, B., Lorenzini, C. A., Baldi, E., Tassoni, G., & Bucherelli, C. (1999). Auditory
thalamus, dorsal hippocampus, basolateral amygdala, and perirhinal cortex role
in the consolidation of conditioned freezing to context and to acoustic stimulus
in the rat. The Journal of Neuroscience, 19, 9570-9578.
Santini, E., Ge, H., Ren, K., de Ortiz, S. P., & Quirk, G. J. (2004). Consolidation of fear
extinction requires protein synthesis in the medial prefrontal cortex. The Journal
of Neuroscience, 24, 5704-5710.
Sarter, M., Bruno, J. P., & Berntson, G. G. (2001). Psychotogenic properties of
benzodiazepine receptor inverse agonists. Psychopharmacology, 156, 1-13.
Scaife, J. C., Langley, R. W., Bradshaw, C. M., Szabadi, E. (2005). Diazepam
suppressed the acquisition but not the expression of ‘fear-potentiation’ of the
acoustic startle response in man. Journal of Psychopharmacology, 19, 347-356.
Schafe, G. E., Atkins, C. M., Swank, M. W., Bauer, E. P., Sweatt, J. D., & LeDoux, J.
E. (2000). Activation of ERK/MAP kinase in the amygdala is required for
memory consolidation of Pavlovian fear conditioning. The Journal of
Neuroscience, 20, 8177-8187.
Schafe, G. E., & LeDoux, J. E. (2000). Memory consolidation of auditory Pavlovian
fear conditioning requires protein synthesis and protein kinase A in the
amygdala. The Journal of Neuroscience, 20:RC96, 1-5.
Schiller, D., Cain, C. K., Curley, N. G., Schwartz, J. S., Stern, S. A., LeDoux, J. E., &
Phelps, E. A. (2008). Evidence for recovery of fear following immediate
extinction in rats and humans. Learning & Memory, 15, 394-402.
214
Schull, J. A. (1979). A conditioned opponent theory of Pavlovian conditioning and
habituation. In Bower, G. (Ed.), The Psychology of Learning and Motivation,
57-90. New York: Academic Press
Shinnick-Gallagher, P., McKernan, M. G., Xie, J., & Zinebi, F. (2003). L-type voltage-
gated calcium channels are involved in the in vivo and in vitro expression of fear
conditioning. Annals of the New York Academy of Sciences, 985, 135-149.
Sierra-Mercado, D., Corcoran, K. A., Lebron-Milad, K., & Quirk, G. J. (2006).
Inactivation of the ventromedial prefrontal cortex reduces expression of
conditioned fear and impairs subsequent recall of extinction. European Journal
of Neuroscience, 24, 1751-1758.
Solomon, R. L. & Corbit, J. D. (1974). An opponent-process theory of motivation: I.
Temporal dynamics of affect. Psychological Review, 81, 119-145.
Soltysik, S. S., Wolfe, G. E., Nicholas, T., Wilson, W. J., & Garcia-Sanchez, J. L.
(1983). Blocking of inhibitory conditioning within a serial conditioned stimulus-
conditioned inhibitor compound: maintenance of acquired behavior without an
unconditioned stimulus. Learning and Motivation, 14, 1-29.
Sotres-Bayon, F., Bush, D. E. A., & LeDoux, J. E. (2007). Acquisition of fear extinction
requires activation of NR2B-containing NMDA receptors in the lateral
amygdala. Neuropsychopharmacology, 32, 1929-1940.
Sotres-Bayon, F., Cain, C. K., & LeDoux, J. E. (2006). Brain mechanisms of fear
extinction: Historical perspectives on the contribution of prefrontal cortex.
Biological Psychiatry, 60, 329-336.
Sotres-Bayon, F., Diaz-Mataix, L., Bush, D. E. A., & LeDoux, J. E. (in press).
Dissociable roles for the ventromedial prefrontal cortex and amygdala in fear
extinction: NR2B contribution. Cerebral Cortex.
215
Stevenson, C. W., Halliday, D. M., Marsden, C. A., & Mason, R. (2007). Systemic
administration of the benzodiazepine receptor partial inverse agonist FG-7142
disrupts corticolimbic network interactions. Synapse, 61, 646-663.
Stork, O., Ji, F., & Obata, K. (2002). Reduction of extracellular GABA in the mouse
amygdala during and following confrontation with a conditioned fear stimulus.
Neuroscience Letters, 327, 138-142.
Stout, S. C., Arcediano, P., Escobar, M., & Miller, R. R. (2003). Overshadowing as a
function of trial number: Dynamics of first-and second-order comparator effects.
Learning & Behavior, 31, 85-97.
Stout, S. C. & Miller, R. R. (2007). Sometimes-competing retrieval (SOCR): A
formalization of the comparator hypothesis. Psychological Review, 114, 759-
783.
Suzuki, A., Josselyn, S. A., Frankland, P. W., Masushige, S., Silva, A. J., & Kida, S.
(2004). Memory reconsolidation and extinction have distinct temporal and
biochemical signatures. The Journal of Neuroscience, 24, 4787-4795.
Szinyei, C., Heinbockel, T., Montagne, J., & Pape, H. (2000). Putative cortical and
thalamic inputs elicit convergent excitation in a population of GABAergic
interneurons of the lateral amygdala. The Journal of Neuroscience, 20, 8909-
8195.
Szinyei, C., Narayanan, R. T., & Pape, H. (2007). Plasticity of inhibitory synaptic
network interactions in the lateral amygdala upon fear conditioning in mice.
European Journal of Neuroscience, 25, 1205-1211.
Tam, S. & Roth, R. H. (1985). Selective increase in dopamine metabolism in the
prefrontal cortex by the anxiogenic beta-carboline FG 7142. Biochemical
Pharmacology, 34, 1595-1598.
216
Thiebot, M. H., Dangoumau, L., Richard, G. & Puech, A. J. (1991). Safety signal
withdrawal: A behavioural paradigm sensitive to both “anxiolytic” and
“anxiogenic” drugs under identical experimental conditions.
Psychopharmacology, 103, 415-424.
Van Hamme, L. J. & Wasserman, E. A. (1994). Cue competition in causality
judgements: The role of nonpresentation of compound stimulus elements.
Learning and Motivation, 25, 127-151.
Venton, B. J., Robinson, T. E., Kennedy, R. T., & Maren, S. (2006). Dynamic amino
acid increases in the basolateral amygdala during acquisition and expression of
conditioned fear. European Journal of Neuroscience, 23, 3391-3398.
Vidal-Gonzalez, I., Vidal-Gonzalez, B., Rauch, S. L., & Quirk, G. J. (2006).
Microstimulation reveals opposing influences of prelimbic and infralimbic
cortex on the expression of conditioned fear. Learning & Memory, 13, 728-733.
Wagner, A. R. (1981). SOP: A model of automatic memory processing in animal
behavior. In Spear, N. E. & Miller, R. R. (Eds.) Information Processing in
Animals: Memory Mechanisms, 5-47. Hillsdale, New Jersey: Erlbaum.
Watson, J. B. & Watson, R. R. (1921). Studies in infant psychology. Scientific Monthly,
13, 493-515.
Westbrook, R. F., Iordanova, M., McNally, G., Richardson, R., & Harris, J. A. (2002).
Reinstatement of fear to an extinguished conditioning stimulus: Two roles for
context. Journal of Experimental Psychology: Animal Behavior Processes, 28,
97-110.
Wilensky, A. E., Schafe, G. E., Kristensen, M. P., & LeDoux, J. E. (2006). Rethinking
the fear circuit: The central nucleus of the amygdala is required for the
217
acquisition, consolidation, and expression of Pavlovian fear conditioning. The
Journal of Neuroscience, 26, 12387-12396.
Wilensky, A. E., Schafe, G. E., & LeDoux, J. E. (1999). Functional inactivation of the
amygdala before but not after auditory fear conditioning prevents memory
formation. The Journal of Neuroscience, 19:RC48, 1-5.
Wu, Z., Thomas, S. A., Villacres, E. C., Xia, Z., Simmons, M. L., Chavkin, C.,
Palmiter, R. D., & Storm, D. R. (1995). Altered behavior and long-term
potentiation in type I adenylyl cyclase mutant mice. Proceedings of the National
Academy of Sciences of the United States of America, 92, 220-224.
Yeh, S., Mao, S., Lin, H., & Gean, P. (2006). Synaptic expression of glutamate receptor
after encoding of fear memory in the rat amygdala. Molecular Pharmacology,
69, 299-308.
Young, S. L. & Fanselow, M. S. (1992). Associative regulation of Pavlovian fear
conditioning: Unconditioned stimulus intensity, incentive shifts, and latent
inhibition. Journal of Experimental Psychology: Animal Behavior Processes, 18,
400-413.
Zimmer-Hart, C. L. & Rescorla, R. A. (1974). Extinction of Pavlovian conditioned
inhibition. Journal of Comparative and Physiological Psychology, 86, 837-845.
Zimmerman, J. M., Rabinak, C. A., McLachlan, I. G., & Maren, S. (2007). The central
nucleus of the amygdala is essential for acquiring and expressing conditional
fear after overtraining. Learning & Memory, 14, 634-644.
Zinebi, F., Xie, J., Liu, J., Russell, R. T., Gallagher, J. P., McKernan, M. G., &
Shinnick-Gallagher, P. (2003). NMDA currents and receptor protein are
downregulated in the amygdala during maintenance of fear memory. The
Journal of Neuroscience, 23, 10283-10291.
218 Appendix 1: Experiment 1 Raw Data Stage I conditioning Tone Control Day 1 Day 3 Day 5 rat pre-CS CS pre-CS CS context CS 1 0 13.3 0 53.3 0 53.3 2 0 20 0 93.3 31.8 73.3 9 0 33.3 0 13.3 0 73.3 10 0 0 40 93.3 0 73.3 18 0 13.3 20 33.3 36.4 20 19 0 60 100 86.7 4.5 53.3 25 0 20 46.7 66.7 9.1 66.7 31 0 26.7 0.0 66.7 0.0 26.7 33 0 0 6.7 60.0 0.0 33.3 35 0 53.3 54.5 86.7 40.9 86.7 39 9.1 33.3 81.8 73.3 0.0 73.3 44 0 0 27.3 53.3 36.4 26.7 45 9.1 33.3 81.8 33.3 31.8 26.7 mean 1.4 23.6 35.3 62.6 14.7 52.8 sem 0.9 5.3 9.8 6.9 4.8 6.5
Over - 2 rat pre-CS CS pre-CS CS context CS 5 0 20 0 46.7 0 40 6 13.3 40 66.7 93.3 0 80 15 6.7 13.3 26.7 80 0 85.7 16 0 6.7 46.7 73.3 4.5 42.9 21 0 14.3 73.3 0 40.9 73.3 22 6.7 7.1 33.3 46.7 31.8 40 28 13.3 21.4 0 40 0 46.7 29 0 7.1 33.3 60 4.5 6.7 36 0 0 18.2 66.7 9.1 46.7 37 0 6.7 4.5 33.3 0.0 26.7 47 0 20 27.3 60.0 31.8 46.7 48 0 13.3 0.0 73.3 0.0 53.3 mean 3.3 14.2 27.5 56.1 10.2 49.1 sem 1.5 3.0 7.3 7.2 4.4 6.4
219 Over - 4 Day 1 Day 3 Day 5 rat pre-CS CS pre-CS CS context CS 3 0 0 60 93.3 13.6 93.3 4 0 6.7 60 86.7 0 20 13 0 53.3 0 73.3 9.1 20 14 0 0 20 66.7 0 73.3 17 0 33.3 0 40 13.6 20 26 0 35.7 80 80 45.5 46.7 27 0 21.4 26.7 53.3 18.2 20 30 0 13.3 9.1 40.0 9.1 20.0 32 0 13.3 0.0 86.7 6.7 53.3 40 0 53.3 0.0 93.3 0.0 100.0 41 4.5 0 9.1 53.3 18.2 40.0 46 13.6 40 4.5 40.0 4.5 53.3 mean 1.5 22.5 22.5 67.2 11.5 46.7 sem 1.2 5.8 8.2 6.1 3.6 8.4
Over - 8 rat pre-CS CS pre-CS CS context CS 7 26.7 60 6.7 33.3 0 57.1 8 0 13.3 33.3 46.7 68.2 0 11 0 13.3 33.3 80 0 26.7 12 0 6.7 20 73.3 0 53.3 20 6.7 0 46.7 80 4.5 40 23 0 0 53.3 93.3 40.9 73.3 24 0 46.7 93.3 93.3 40.9 6.7 34 0 40 0.0 33.3 13.6 46.7 38 0 20 0.0 33.3 0.0 46.7 42 0 66.7 9.1 20.0 13.6 46.7 43 0 33.3 13.6 86.7 40.9 33.3 mean 3.0 27.3 28.1 61.2 20.2 39.1 sem 2.4 7.1 8.5 8.4 7.1 6.5
Flashing light
Control Day 2 Day 4 Day 6 rat pre-CS CS pre-CS CS context CS 1 6.7 26.7 40 46.7 0 73.3 2 26.7 80 46.7 93.3 13.3 86.7 9 6.7 0 0 20 0 86.7 10 6.7 33.3 66.7 40 0 73.3 18 6.7 0 20 42.9 13.3 0 19 13.3 86.7 20 92.9 6.7 53.3 25 0 0 20 57.1 0 66.7 31 0 46.7 0.0 13.3 0.0 80.0 33 0 6.7 0.0 60.0 10.0 53.3 35 4.5 33.3 95.5 93.3 10.0 53.3 39 4.5 20 27.3 33.3 4.5 60.0 44 27.3 0 81.8 33.3 4.5 26.7 45 59.1 100 72.7 46.7 72.7 93.3 mean 12.5 33.3 37.7 51.8 10.4 62.0 sem 4.6 9.8 9.0 7.5 5.4 7.2
220 Over - 2 Day 2 Day 4 Day 6 rat pre-CS CS pre-CS CS context CS 5 13.3 66.7 20 33.3 0 66.7 6 40 66.7 6.7 80 0 73.3 15 0 26.7 33.3 53.3 6.7 66.7 16 0 0 13.3 0 0 60 21 80 26.7 40 13.3 0 26.7 22 53.3 100 46.7 46.7 26.7 0 28 13.3 0 13.3 53.3 0 33.3 29 0 0 20 20 0 13.3 36 9.1 46.7 13.6 20.0 3.3 20.0 37 0 13.3 0.0 13.3 0.0 53.3 47 13.6 20 40.9 40.0 9.1 80.0 48 0 53.3 0.0 53.3 0.0 66.7 mean 18.6 35.0 20.7 35.5 3.8 46.7 sem 7.5 9.2 4.6 6.6 2.3 7.7
Over - 4 rat pre-CS CS pre-CS CS context CS 3 26.7 93.3 93.3 80 6.7 66.7 4 0 13.3 13.3 86.7 0 80 13 6.7 80 6.7 66.7 0 6.7 14 6.7 26.7 20 86.7 0 6.7 17 6.7 7.1 6.7 0 20 13.3 26 93.3 13.3 60 86.7 13.3 80 27 6.7 0 33.3 40 0 26.7 30 4.5 40 0.0 46.7 6.7 53.3 32 0 40 43.3 26.7 3.3 6.7 40 0 33.3 22.7 80.0 0.0 93.3 41 18.2 20 0.0 26.7 13.6 46.7 46 0 20 13.6 53.3 4.5 40.0 mean 14.1 32.3 26.1 56.7 5.7 43.3 sem 7.6 8.2 8.0 8.4 1.9 9.2
Over - 8 rat pre-CS CS pre-CS CS context CS 7 20 6.7 6.7 33.3 0 33.3 8 6.7 20 46.7 6.7 13.3 6.7 11 13.3 33.3 6.7 60 0 60 12 46.7 26.7 13.3 80 0 33.3 20 13.3 93.3 20 86.7 0 53.3 23 26.7 21.4 100 13.3 66.7 86.7 24 0 0 30.8 93.3 20 20 34 22.7 33.3 9.1 80.0 23.3 53.3 38 13.6 13.3 4.5 46.7 0.0 46.7 42 86.4 0 45.5 73.3 4.5 80.0 43 13.6 60 40.9 60.0 27.3 66.7 mean 23.9 28.0 29.5 57.6 14.1 49.1 sem 7.2 8.4 8.6 8.9 6.1 7.4
221 Stage II conditioning
Over - 2 rat context CS1 CS2 5 0 73.3 46.7 6 0 73.3 60 15 0 60 73.3 16 0 53.3 6.7 21 20 73.3 64.3 22 6.7 40 50 28 0 60 64.3 29 6.7 60 42.9 36 9.1 93.3 46.7 37 0 66.7 40 47 20 20 6.7 48 3.3 73.3 60 mean 5.5 62.2 46.8 sem 2.2 5.4 6.1
Over - 4 rat context CS1 CS2 CS3 CS4 3 0 93.3 80 93.3 73.3 4 0 100 80 60 13.3 13 0 40 33.3 0 20 14 0 80 86.7 46.7 20 17 6.7 0 20 13.371.4 26 6.7 86.7 66.7 53.3 50 27 0 33.3 26.7 20 35.7 30 13.6 86.7 46.7 26.7 46.7 32 0 73.3 33.3 40 46.7 40 0 80 93.3 46.7 20 41 0 20 20 26.726.7 46 10 40 53.3 40 60 mean 3.1 61.1 53.3 38.9 40.3 sem 1.4 9.5 7.8 7.0 6.0
Over - 8 rat context CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 7 0 26.7 13.3 2013.3 33.3 46.7 13.3 40 8 0 26.7 26.7 40 6.7 26.7 26.7 0 26.7 11 0 80 46.7 13.326.7 40 20 20 20 12 0 73.3 60 60 40 46.7 33.3 20 40 20 0 80 33.3 20 26.7 6.7 0 20 33.3 23 0 66.7 93.3 80.0 46.7 60.0 66.7 73.3 20.0 24 0 60 26.7 60 13.3 0 0 0 26.7 34 13.3 53.3 33.3 46.7 40 46.7 40 33.3 13.3 38 0 73.3 33.3 26.7 33.3 13.3 6.7 0 26.7 42 13.6 86.7 60 21.4 46.7 0 33.3 26.7 40 43 54.5 100 20 14.3 33.3 40 46.7 40 46.7 mean 7.4 66.1 40.6 36.6 29.7 28.5 29.1 22.4 30.3 sem 5.0 7.0 6.9 6.8 4.2 6.2 6.4 6.5 3.2
222 Test
Control rat context CS1 CS2 CS3 CS4 CS mean 1 6.7 73.3 33.333.3 6.7 36.7 2 11.1 73.3 80 86.7 86.7 81.7 9 8.9 66.7 86.773.3 80 76.7 10 0 53.3 73.360 40 56.7 18 46.7 66.7 60 40 26.7 48.4 19 22.2 26.7 80 73.3 53.3 58.3 25 0 26.7 2013.3 40 25.0 31 0 46.7 4046.7 33.3 41.7 33 0 60 86.746.7 40 58.4 35 45.5 66.7 60 0 73.3 50.0 39 10 80 86.7 60 66.7 73.4 44 4.5 26.7 40 26.7 13.3 26.7 45 16.7 66.7 93.3 53.3 53.3 66.7 mean 13.3 56.4 64.647.2 47.2 53.8 sem 4.5 5.3 6.76.9 6.9 5.0
Over - 2 rat context CS1 CS2 CS3 CS4 CS mean 5 4.4 26.7 40 60 20 36.7 6 11.1 46.7 13.3 33.3 26.7 30.0 15 4.4 80 80 20 53.3 58.3 16 13.3 26.7 0 0 20 11.7 21 24.4 80 86.7 40 40 61.7 22 17.8 73.3 46.7 13.3 13.3 36.7 28 0 80 6053.3 33.3 56.7 29 11.1 20 6.7 6.7 13.3 11.7 36 10 53.3 73.373.3 6.7 51.7 37 4.5 40 40 33.3 20 33.3 47 18.2 6.7 26.7 20 26.7 20.0 48 3.3 33.3 26.753.3 53.3 41.7 mean 10.2 47.2 41.733.9 27.2 37.5 sem 2.1 7.5 8.36.6 4.4 5.0
223 Over - 4 rat context CS1 CS2 CS3 CS4 CS mean 3 44.4 60 73.3 66.7 33.3 58.3 4 24.5 0 33.3 20 6.7 15.0 13 15.6 60 60 0 20 35.0 14 15.5 33.3 40 40 33.3 36.7 17 15.6 40 40 20 13.3 28.3 26 0 80 100 80 86.7 86.7 27 4.4 13.3 60 13.3 33.3 30.0 30 9.1 40 66.7 46.7 20 43.4 32 45.5 73.3 80 80 66.7 75.0 40 0 66.7 86.7 73.3 13.3 60.0 41 9.1 26.7 13.3 53.3 53.3 36.7 46 3.3 40 26.7 6.7 13.3 21.7 mean 15.6 44.4 56.7 41.7 32.8 43.9 sem 4.5 7.0 7.6 8.5 7.1 6.3
Over - 8 rat context CS1 CS2 CS3 CS4 CS mean 7 0 66.7 93.3 73.3 60 73.3 8 11.1 26.7 6.7 6.7 40 20.0 11 6.7 60 66.7 26.746.7 50.0 12 0 26.7 53.3 4046.7 41.7 20 20 86.7 100 93.3 60 85.0 23 15.6 80 93.3 53.3 40 66.7 24 44.4 20 80 73.3 86.7 65.0 34 31.8 46.7 86.7 80 86.7 75.0 38 4.5 53.3 46.7 26.7 20 36.7 42 43.3 46.7 20 40 33.3 35.0 43 23.3 100 93.3 73.3 53.3 80.0 mean 18.2 55.8 67.3 53.3 52.1 57.1 sem 4.8 7.8 9.6 8.2 6.2 6.5
Statistical Analyses
Stage I: context vs. CS freezing on Days 5 and 6.
Number of Groups: 1 Number of Measurements: 4
Number of subjects in... Group 1: 48
224 Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 1 1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 30.057 Measurement 1 2 3 4 Mean 14.054 8.417 47.202 50.556 SD 17.369 15.012 24.041 27.496 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 28449.590 47 605.310 ------Within ------W1 68018.492 1 68018.492 91.160 Error 35068.868 47 746.146 ------
Stage II: Context freezing vs. CS freezing
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 35
225 Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 25.838 Measurement 1 2 Mean 5.263 46.414 SD 10.461 18.190 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 7241.613 34 212.989 ------Within ------W1 29634.687 1 29634.687 130.366 Error 7728.864 34 227.320 ------
Stage II: Evidence for summation Comparison of mean CS freezing on Days 5 and 6 to freezing to CS1 on Day 7
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 35
226 Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 54.373 Measurement 1 2 Mean 45.703 63.043 SD 21.628 24.959 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 26149.621 34 769.106 ------Within ------W1 5261.823 1 5261.823 16.359 Error 10936.080 34 321.649 ------
Comparison of freezing to tone CS on Day 5 to freezing CS1 on Day 7
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 35
227 Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 54.080 Measurement 1 2 Mean 45.117 63.043 SD 24.301 24.959 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 26465.462 34 778.396 ------Within ------W1 5623.297 1 5623.297 12.924 Error 14794.013 34 435.118 ------
Comparison of freezing to flashing light CS on Day 6 to freezing to CS1 on Day 7
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 35
228 Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 54.666 Measurement 1 2 Mean 46.289 63.043 SD 27.148 24.959 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 32497.308 34 955.803 ------Within ------W1 4912.357 1 4912.357 12.154 Error 13741.673 34 404.167 ------
Stage II: Freezing to first 2 CS presentations
Number of Groups: 3 Number of Measurements: 2
Number of subjects in... Group 1 (Over - 2): 12 Group 2 (Over - 4): 12 Group 3 (Over - 8): 11
229 Between contrast coefficients Contrast Group... 1 2 3 B1 2 -1 -1 B2 -1 2 -1 B3 -1 -1 2 B4 1 -1 0 B5 1 0 -1 B6 0 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 54.504 Measurement 1 2 Mean 62.208 46.800 SD 18.703 21.215
Group 2 Overall Mean: 57.221 Measurement 1 2 Mean 61.108 53.333 SD 32.766 27.119
Group 3 Overall Mean: 53.332 Measurement 1 2 Mean 66.064 40.600 SD 23.173 22.985
230 Means and SDs averaged across groups Measurement 1 2 Mean 63.127 46.911 SD 25.634 23.930
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 9.397 1 9.397 0.010 B2 171.935 1 171.935 0.175 B3 96.614 1 96.614 0.098 B4 88.563 1 88.563 0.090 B5 15.776 1 15.776 0.016 B6 173.602 1 173.602 0.177 Error 31403.192 32 981.350 ------Within ------W1 4593.848 1 4593.848 18.496 B1W1 5.778 1 5.778 0.023 B2W1 631.632 1 631.632 2.543 B3W1 725.742 1 725.742 2.922 B4W1 174.803 1 174.803 0.704 B5W1 290.139 1 290.139 1.168 B6W1 897.852 1 897.852 3.615 Error 7947.919 32 248.372 ------Stage II: Freezing to first 4 CS presentations
Number of Groups: 2 Number of Measurements: 4
231 Number of subjects in... Group 1 (Over - 4): 12 Group 2 (Over - 8): 11
Between contrast coefficients Contrast Group... 1 2 B1 1 -1 Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 -3 -1 1 3
Means and Standard Deviations Group 1 Overall Mean: 48.413 Measurement 1 2 3 4 Mean 61.108 53.333 38.892 40.317 SD 32.766 27.119 24.420 20.871
Group 2 Overall Mean: 43.236 Measurement 1 2 3 4 Mean 66.064 40.600 36.582 29.700 SD 23.173 22.985 22.389 13.788
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 63.586 46.967 37.737 35.008 SD 28.602 25.235 23.474 17.852 ------
232 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 615.060 1 615.060 0.512 Error 25236.479 21 1201.737 ------Within ------W1 10351.036 1 10351.036 17.718 B1W1 377.962 1 377.962 0.647 Error 12268.186 21 584.199 ------
Stage II: All 8 CS presentations for group Over - 8
Number of Groups: 1 Number of Measurements: 8
Number of subjects in... Group 1: 11
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 6 7 8 W1 -7 -5 -3 -1 1 3 5 7
Means and Standard Deviations Group 1 Overall Mean: 35.408 Measurement 1 2 3 4 5 6 7 8 Mean 66.064 40.600 36.582 29.700 28.491 29.100 22.418 30.309 SD 23.173 22.985 22.389 13.788 20.679 21.145 21.548 10.489
233 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 12237.311 10 1223.731 ------Within ------W1 8715.680 1 8715.680 25.960 Error 3357.408 10 335.741 ------
Test: Test for possible differences in context freezing
Number of Groups: 4 Number of Measurements: 1
Number of subjects in... Group 1 (Control): 13 Group 2 (Over - 2): 12 Group 3 (Over - 4): 12 Group 4 (Over - 8): 11
234 Between contrast coefficients Contrast Group... 1 2 3 4 B1 3 -1 -1 -1 B2 1 -1 0 0 B3 1 0 -1 0 B4 1 0 0 -1 B5 0 2 -1 -1 B6 0 1 -1 0 B7 0 1 0 -1 B8 0 0 1 -1 B9 0 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 13.254 Measurement 1 Mean 13.254 SD 16.090
Group 2 Overall Mean: 10.208 Measurement 1 Mean 10.208 SD 7.299
Group 3 Overall Mean: 15.583 Measurement 1 Mean 15.583 SD 15.523
Group 4 Overall Mean: 18.245 Measurement 1 Mean 18.245 SD 16.032
235 Means and SDs averaged across groups Measurement 1 Mean 14.323 SD 14.233 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between ------B1 19.245 1 19.245 0.095 B2 57.877 1 57.877 0.286 B3 33.861 1 33.861 0.167 B4 148.459 1 148.459 0.733 B5 354.400 1 354.400 1.749 B6 173.344 1 173.344 0.856 B7 370.721 1 370.721 1.830 B8 40.673 1 40.673 0.201 B9 215.865 1 215.865 1.066 Error 8913.725 44 202.585 ------
Test: Comparison of context freezing vs mean CS freezing
Number of Groups: 4 Number of Measurements: 5
236 Number of subjects in... Group 1 (Control): 13 Group 2 (Over - 2): 12 Group 3 (Over - 4): 12 Group 4 (Over - 8): 11
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 W1 4 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 45.729 Measurement 1 2 3 4 5 Mean 13.254 56.423 64.615 47.177 47.177 SD 16.090 18.964 24.256 24.716 24.716
Group 2 Overall Mean: 32.040 Measurement 1 2 3 4 5 Mean 10.208 47.225 41.675 33.875 27.217 SD 7.299 25.958 28.725 22.812 15.154
Group 3 Overall Mean: 38.225 Measurement 1 2 3 4 5 Mean 15.583 44.442 56.667 41.667 32.767 SD 15.523 24.345 26.292 29.282 24.541
Group 4 Overall Mean: 49.349 Measurement 1 2 3 4 5 Mean 18.245 55.773 67.273 53.327 52.127 SD 16.032 26.034 31.883 27.301 20.629
237 Means and SDs averaged across groups Measurement 1 2 3 4 5 Mean 14.323 50.966 57.557 44.011 39.822 SD 14.233 23.848 27.760 26.085 21.709 ------
Analysis of Variance Summary Table
Source SS df MS F ------Between 58726.840 44 1334.701 ------Within ------W1 43629.865 1 43629.865 103.586 Error 18532.547 44 421.194 ------
Test: CS freezing
Number of Groups: 4 Number of Measurements: 4
Number of subjects in... Group 1 (Control): 13 Group 2 (Over - 2): 12 Group 3 (Over - 4): 12 Group 4 (Over - 8): 11
238 Between contrast coefficients Contrast Group... 1 2 3 4 B1 0 -1 0 1 B2 1 -1 0 0 B3 1 0 -1 0 B4 1 0 0 -1
Means and Standard Deviations Group 1 Overall Mean: 53.848 Measurement 1 2 3 4 Mean 56.423 64.615 47.177 47.177 SD 18.964 24.256 24.716 24.716 Group 2 Overall Mean: 37.498 Measurement 1 2 3 4 Mean 47.225 41.675 33.875 27.217 SD 25.958 28.725 22.812 15.154
Group 3 Overall Mean: 43.885 Measurement 1 2 3 4 Mean 44.442 56.667 41.667 32.767 SD 24.345 26.292 29.282 24.541
Group 4 Overall Mean: 57.125 Measurement 1 2 3 4 Mean 55.773 67.273 53.327 52.127 SD 26.034 31.883 27.301 20.629
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 50.966 57.557 44.011 39.822 SD 23.848 27.760 26.085 21.709 ------
239 Analysis of Variance Summary Table
Source SS df MS F ------Between ------B1 8843.366 1 8843.366 5.693 B2 6672.500 1 6672.500 4.296 B3 2477.395 1 2477.395 1.595 B4 255.928 1 255.928 0.165 Error 68345.662 44 1553.310 ------
240 Appendix 2: Experiment 2 Raw Data Stage I conditioning Tone Control Day 1 Day 3 Day 5 rat pre-CS CS pre-CS CS context CS 1 40 33.3 33.3 40 0 46.7 6 0 66.7 13.3 73.3 4.5 13.3 10 46.7 20 66.7 80 0 53.3 13 6.7 53.3 13.3 46.7 0 73.3 16 26.7 13.3 26.7 93.3 54.5 60 17 0 6.7 0.0 86.7 22.7 73.3 22 53.3 20 60.0 93.3 27.3 60 27 0 6.7 6.7 60.0 33.3 13.3 28 0 6.7 20.0 86.7 3.3 46.7 mean 19.3 25.2 26.7 73.3 16.2 48.9 sem 6.8 6.6 7.0 6.0 5.96.7
Over - Veh rat pre-CS CS pre-CS CS context CS 4 86.7 60 33.3 73.3 45.5 73.3 5 0 6.7 13.3 46.7 4.5 40 7 13.3 66.7 6.7 46.7 0 26.7 12 0 13.3 66.7 66.7 4.5 53.3 14 6.7 13.3 6.7 53.3 0 66.7 15 6.7 33.3 40 53.3 9.1 86.7 18 0 33.3 13.3 46.7 0 20 20 0 20 33.3 0.0 0 33.3 21 33.3 40 86.7 60.0 4.5 60 24 13.3 20 60.0 80.0 10 46.7 26 0 53.3 13.3 60.0 16.7 6.7 mean 14.5 32.7 33.9 53.3 8.6 46.7 sem 7.8 6.1 8.2 6.3 4.07.3
Over – FG7142 rat pre-CS CS pre-CS CS context CS 2 0 40 0 93.3 9.1 46.7 3 6.7 33.3 20 86.7 0 40 8 20 13.3 53.3 80 0 80 9 0 13.3 40 20 0 33.3 11 0 0 60 66.7 0 73.3 19 6.7 66.7 0.0 60.0 4.5 33.3 23 0 66.7 46.7 33.3 3.3 53.3 25 0 33.3 66.7 93.3 80 53.3 29 66.7 33.3 13.3 40.0 30 20 30 0 13.3 33.3 80.0 43.3 53.3 mean 10.0 31.3 33.3 65.3 17.0 48.7 sem 6.6 7.1 7.6 8.3 8.45.8
241
Flashing light Control Day 2 Day 4 Day 6 rat pre-CS CS pre-CS CS context CS 1 0 33.3 26.7 73.3 0 46.7 6 13.3 46.7 33.3 60 0 60 10 0 53.3 13.3 86.7 0 86.7 13 40 46.7 0 100 33.3 73.3 16 0 0 26.7 46.7 26.7 66.7 17 0 26.7 40.0 93.3 0 93.3 22 0 40 33.3 66.7 20 46.7 27 0 13.3 40.0 66.7 35 33.3 28 13.3 6.7 86.7 80.0 0 73.3 mean 7.4 29.6 33.3 74.8 12.8 64.4 sem 4.1 5.8 7.2 5.1 4.7 5.9
Over - Veh rat pre-CS CS pre-CS CS context CS 4 0 13.3 33.3 86.7 0 86.7 5 0 40 20 33.3 0 46.7 7 13.3 20 0 13.3 0 20 12 0 6.7 86.7 86.7 13.3 86.7 14 0 53.3 86.7 80 0 86.7 15 6.7 40 13.3 13.3 0 86.7 18 20 80 6.7 60.0 0 60 20 13.3 46.7 86.7 33.3 6.7 73.3 21 6.7 46.7 60.0 60.0 6.7 46.7 24 0 20 0.0 66.7 0 66.7 26 6.7 46.7 0.0 40.0 25 26.7 mean 6.1 37.6 35.8 52.1 4.7 62.4 sem 2.1 6.4 11.2 8.2 2.4 7.4
Over - FG7142 rat pre-CS CS pre-CS CS context CS 2 0 60 6.7 80 6.7 40 3 13.3 26.7 26.7 6.7 0 53.3 8 6.7 93.3 20 66.7 0 93.3 9 0 0 26.7 93.3 0 13.3 11 0 20 13.3 66.7 0 60 19 13.3 40 13.3 73.3 0 13.3 23 53.3 66.7 0.0 20.0 0 33.3 25 66.7 66.7 86.7 100.0 80 86.7 29 13.3 26.7 46.7 53.3 10 33.3 30 26.7 40 66.7 66.7 0 73.3 mean 19.3 44.0 30.7 62.7 9.7 50.0 sem 7.3 8.7 8.8 9.3 7.9 9.0
242
Stage II conditioning Over - Veh Over - FG7142 rat context CS1 CS2 rat context CS1 CS2 4 93.3 53.3 2 86.7 93.3 5 0 26.76.7 3 26.7 46.7 7 0 60 20 8 0 86.773.3 12 0 26.713.3 9 0 6.733.3 14 0 33.36.7 11 0 80 60 15 0 93.326.7 19 0 60.060.0 18 0 4033.3 23 5 53.313.3 20 5 86.746.7 25 15 86.746.7 21 15 60 46.7 29 40 80.026.7 24 5 80 80 30 3.3 60.053.3 26 10 40 60 mean 7.9 62.7 50.7 mean 3.5 58.235.8 sem 4.4 8.7 7.3 sem 1.6 8.07.1
Test Control rat context CS1 CS2 CS3 CS4 CS mean 1 0 46.773.3 66.7 66.7 63.4 6 26.7 86.7 73.3 100 40 75.0 10 0 80 53.3 80 53.3 66.7 13 0 73.386.7 80 66.7 76.7 16 33.3 20 53.3 73.3 60 51.7 17 0 73.393.3 93.3 53.3 78.3 22 40 86.7 66.7 80 80 78.4 27 0 86.793.3 86.7 66.7 83.4 28 15 60 93.3 100 53.3 76.7 mean 12.8 68.276.3 84.4 60.0 72.2 sem 5.0 6.8 4.9 3.5 3.5 3.0
Over - Veh rat context CS1 CS2 CS3 CS4 CS mean 4 0 73.386.7 93.3 100 88.3 5 0 73.3 40 13.3 20 36.7 7 6.7 20 26.7 53.3 33.3 33.3 12 0 53.373.3 33.3 53.3 53.3 14 0 13.366.7 20 40 35.0 15 13.3 46.7 73.3 46.7 53.3 55.0 18 6.7 53.3 26.7 20 20 30.0 20 6.7 26.7 80 53.3 20 45.0 21 13.3 13.3 73.3 93.3 13.3 48.3 24 15 53.3 86.7 66.7 60 66.7 26 0 6.7 26.7 33.3 40 26.7 mean 5.6 39.4 60.0 47.9 41.2 47.1 sem 1.8 7.3 7.5 8.4 7.6 5.5
243 Over - FG7142 rat context CS1 CS2 CS3 CS4 CS mean 2 0 66.7 86.7 73.3 73.3 75.0 3 0 53.3 73.3 53.3 13.3 48.3 8 0 80 86.7 86.7 73.3 81.7 9 0 53.3 60 53.3 40 51.7 11 0 60 86.7 66.7 66.7 70.0 19 0 26.7 53.3 13.3 13.3 26.7 23 26.7 66.7 46.7 46.7 60 55.0 25 40 93.3 60 80 86.7 80.0 29 5 40 66.7 53.3 60 55.0 30 0 66.7 86.7 86.7 100 85.0 mean 7.2 60.7 70.7 61.3 58.7 62.8 sem 4.3 5.7 4.7 6.8 8.7 5.6
Statistical Analyses Stage I: Comparison of freezing to context and freezing to CSs on days 5 and 6.
Number of Groups: 1 Number of Measurements: 4
Number of subjects in... Group 1: 30
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 1 1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 32.338 Measurement 1 2 3 4 Mean 13.687 8.780 47.993 58.890 SD 20.028 17.191 21.110 24.647 ------
244 Analysis of Variance Summary Table Source SS df MS F ------Between 22974.059 29 792.209 ------Within ------W1 53446.302 1 53446.302 81.358 Error 19050.995 29 656.931 ------
Stage II: Comparison of context freezing to mean CS freezing
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 18
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 27.269 Measurement 1 2 Mean 5.461 49.078 SD 9.997 20.829 ------
245 Analysis of Variance Summary Table
Source SS df MS F ------Between 5234.451 17 307.909 ------Within ------W1 17121.723 1 17121.723 75.804 Error 3839.753 17 225.868 ------
Stage II: Test for evidence of summation
Number of Groups: 1 Number of Measurements: 3
Number of subjects in... Group 1: 21
Within contrast coefficients Contrast Measurement... 1 2 3 W1 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 54.816 Measurement 1 2 3 Mean 47.614 56.510 60.324 SD 21.109 26.562 26.450 ------
246 Analysis of Variance Summary Table Source SS df MS F ------Between 23981.251 20 1199.063 ------Within ------W1 955.627 1 955.627 2.089 Error 9150.246 20 457.512 ------
Stage II: CS freezing
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 21
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 51.590 Measurement 1 2 Mean 60.324 42.857 SD 26.450 24.082 ------
247 Analysis of Variance Summary Table Source SS df MS F ------Between 19314.216 20 965.711 ------Within ------W1 3203.387 1 3203.387 10.207 Error 6277.073 20 313.854 ------
Test: Comparison between context freezing and mean CS freezing Number of Groups: 3 Number of Measurements: 5 Number of subjects in... Group 1 (Control): 9 Group 2 (Over - Veh): 11 Group 3 (Over - FG7142): 10
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 W1 4 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 60.331 Measurement 1 2 3 4 5 Mean 12.778 68.156 76.278 84.444 60.000 SD 16.497 22.557 16.368 11.542 11.561
248 Group 2 Overall Mean: 38.813 Measurement 1 2 3 4 5 Mean 5.609 39.382 60.009 47.864 41.200 SD 6.061 24.291 24.755 27.777 25.091
Group 3 Overall Mean: 51.702 Measurement 1 2 3 4 5 Mean 7.170 60.670 70.680 61.330 58.660 SD 14.235 18.967 15.472 22.416 28.782
Means and SDs averaged across groups Measurement 1 2 3 4 5 Mean 8.519 56.069 68.989 64.546 53.287 SD 12.720 22.119 19.650 22.198 23.429 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 24550.243 27 909.268 ------Within ------W1 64968.030 1 64968.030 200.508 Error 8748.464 27 324.017 ------
Test: Between-groups tests for possible differences in freezing to context
Number of Groups: 3 Number of Measurements: 1
249 Number of subjects in... Group 1 (Control): 9 Group 2 (Over - Veh): 11 Group 3 (Over - FG7142): 10
Between contrast coefficients Contrast Group... 1 2 3 B1 2 -1 -1 B2 1 -1 0 B3 1 0 -1 B4 0 1 -1 B5 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 12.778 Measurement 1 Mean 12.778 SD 16.497
Group 2 Overall Mean: 5.609 Measurement 1 Mean 5.609 SD 6.061
Group 3 Overall Mean: 7.170 Measurement 1 Mean 7.170 SD 14.235
Means and SDs averaged across groups Measurement 1 Mean 8.519 SD 12.720
250 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 256.925 1 256.925 1.588 B2 254.381 1 254.381 1.572 B3 148.960 1 148.960 0.921 B4 12.762 1 12.762 0.079 B5 27.204 1 27.204 0.168 Error 4368.506 27 161.797 ------
Test: Between-group comparison of freezing to CSs
Number of Groups: 3 Number of Measurements: 4
Number of subjects in... Group 1 (Control): 9 Group 2 (Over - Veh): 11 Group 3 (Over - FG7142): 10
Between contrast coefficients Contrast Group... 1 2 3 B1 1 -1 0 B2 0 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 -3 -1 1 3
251 Means and Standard Deviations Group 1 Overall Mean: 72.219 Measurement 1 2 3 4 Mean 68.156 76.278 84.444 60.000 SD 22.557 16.368 11.542 11.561
Group 2 Overall Mean: 47.114 Measurement 1 2 3 4 Mean 39.382 60.009 47.864 41.200 SD 24.291 24.755 27.777 25.091
Group 3 Overall Mean: 62.835 Measurement 1 2 3 4 Mean 60.670 70.680 61.330 58.660 SD 18.967 15.472 22.416 28.782
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 56.069 68.989 64.546 53.287 SD 22.119 19.650 22.198 23.429 ------
252 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 12479.972 1 12479.972 11.647 B2 5178.617 1 5178.617 4.833 Error 28930.202 27 1071.489 ------Within ------W1 243.746 1 243.746 0.834 B1W1 22.853 1 22.853 0.078 B2W1 19.774 1 19.774 0.068 Error 7894.897 27 292.404 ------
253 Appendix 3: Experiment 3 Raw Data Stage I conditioning Tone Control Day 1 Day 3 rat context CS context CS 1 0 60 45.5 33.3 7 0 0 18.2 73.3 9 0 0 4.5 66.7 12 0 0 0 80 13 0 26.7 86.4 86.7 19 0 26.7 0.0 73.3 24 0 13.3 40.9 40.0 26 0 13.3 0.0 40.0 30 0 46.7 13.6 53.3 35 0 40 4.5 33.3 38 0 40 4.5 86.7 40 0 46.7 9.1 60.0 mean 0.0 26.1 18.960.6 sem 0.0 6.0 7.6 5.8
Over – 0 mg/kg Day 1 Day 3 rat context CS context CS 2 0 20 4.5 60 6 0 20 9.1 80 10 0 26.7 0 33.3 14 0 0 68.2 40 16 0 20 0 86.7 22 0 40 6.7 33.3 23 0 20 6.7 80.0 27 0 20 0.0 40.0 31 0 26.7 13.6 53.3 32 0 0 0.0 66.7 37 4.5 60 27.3 40.0 mean 0.4 23.0 12.455.8 sem 0.4 5.0 6.1 6.1
Over - 1 mg/kg Day 1 Day 3 rat context CS context CS 3 0 13.3 4.5 33.3 8 0 20 95.5 100 11 0 0 0 40 15 0 26.7 9.1 73.3 18 0 0 0.0 53.3 25 4.5 46.7 0.0 40.0 34 0 13.3 0.0 66.7 39 0 26.7 0.0 46.7 mean 0.6 18.3 13.656.7 sem 0.6 5.5 11.87.9
254
Over - 10 mg/kg Day 1 Day 3 rat context CS context CS 4 0 0 0 33.3 5 0 0 0 80 17 0 53.3 63.6 93.3 20 4.5 66.7 9.1 33.3 21 0 26.7 23.3 86.7 28 0 6.7 0.0 66.7 29 0 13.3 22.7 66.7 33 0 6.7 0.0 40.0 36 0 6.7 0.0 40.0 mean 0.5 20.0 13.2 60.0 sem 0.5 8.1 7.1 7.9
Flashing light Control Day 2 Day 4 rat context CS context CS 1 54.5 80 31.8 0 7 4.5 66.7 31.8 13.3 9 4.5 6.7 0 0 12 0 6.7 0 86.7 13 0 66.7 36.4 93.3 19 13.6 93.3 13.6 60.0 24 0.0 20.0 0.0 26.7 26 0.0 0.0 4.5 33.3 30 13.6 93.3 0.0 53.3 35 0.0 13.3 0.0 100.0 38 0.0 40.0 0.0 86.7 40 4.5 0.0 0.0 53.3 mean 7.9 40.6 9.8 50.6 sem 4.5 10.7 4.3 10.4
Over – 0 mg/kg Day 2 Day 4 rat context CS context CS 2 0 20 4.5 26.7 6 0 26.7 4.5 66.7 10 4.5 6.7 0 33.3 14 4.5 0 54.5 40 16 0 26.7 0 80 22 0.0 13.3 0.0 0.0 23 13.3 66.7 0.0 86.7 27 13.6 0.0 0.0 46.7 31 4.5 13.3 0.0 33.3 32 0.0 0.0 4.5 13.3 37 13.6 26.7 50.0 20.0 mean 4.9 18.2 10.7 40.6 sem 1.8 5.8 6.2 8.3
255
Over - 1 mg/kg Day 2 Day 4 rat context CS context CS 3 0 33.3 9.1 20 8 31.8 80 0 93.3 11 0 46.7 0 33.3 15 13.6 26.7 9.1 46.7 18 0.0 66.7 0.0 80.0 25 4.5 40.0 13.6 26.7 34 0.0 20.0 0.0 40.0 39 0.0 53.3 0.0 80.0 mean 6.2 45.8 4.0 52.5 sem 4.0 7.2 2.0 9.9
Over - 10 mg/kg Day 2 Day 4 rat context CS context CS 1.7 0 0 0 46.7 1.8 0 26.7 0 40 4.8 59.1 6.7 13.6 46.7 5.4 18.2 26.7 9.1 66.7 5.5 13.3 93.3 0.0 53.3 6.8 9.1 13.3 40.9 20.0 7.1 59.1 40.0 22.7 46.7 7.6 0.0 40.0 0.0 86.7 8.3 9.1 26.7 0.0 66.7 mean 18.7 30.4 9.6 52.6 sem 7.9 9.1 4.8 6.3
Stage II conditioning Over – 0 mg/kg rat context CS1 CS2 2 9.1 26.7 0 6 0 33.3 20 10 9.1 40 20 14 27.3 80 33.3 16 0 86.7 80 22 0 53.3 20 23 0 46.7 20 27 4.5 60 20 31 0 66.7 33.3 32 0 21.4 26.7 37 36.4 53.3 53.3 mean 7.9 51.6 29.7 sem 3.8 6.3 6.4
256
Over - 1 mg/kg rat context CS1 CS2 3 0 40 6.7 8 0 86.7 60 11 4.5 60 73.3 15 0 73.3 80 18 0 33.313.3 25 9.1 26.7 0 34 4.5 66.7 53.3 39 0 93.346.7 mean 2.3 60.0 41.7 sem 1.2 8.7 11.0
Over - 10 mg/kg rat context CS1 CS2 1.7 13.6 53.3 40 1.8 0 73.3 13.3 4.8 9.1 93.3 40 5.4 0 66.7 26.7 5.5 0 73.3 26.7 6.8 31.8 60 26.7 7.1 4.5 53.3 13.3 7.6 0 78.6 40 8.3 0 42.9 53.3 mean 6.6 66.1 31.1 sem 3.6 5.1 4.4
Test Control rat context CS1 CS2 CS3 CS4 CS mean 1 4.5 40 40 13.3 26.7 30.0 7 0 93.3 53.3 73.3 33.3 63.3 9 0 60 26.7 26.7 60 43.4 12 0 53.3 73.3 80 60 66.7 13 27.3 53.3 93.3 86.7 73.3 76.7 19 4.5 53.3 73.3 53.3 46.7 56.7 24 0 60 53.3 13.3 66.7 48.3 26 4.5 100 80 60 33.3 68.3 30 0 40 0 26.7 0 16.7 35 0 46.7 66.7 20 80 53.4 38 4.5 33.3 66.7 33.3 53.3 46.7 40 18.2 46.7 66.7 26.7 53.3 48.4 mean 5.3 56.7 57.8 42.8 48.9 51.5 sem 2.5 5.9 7.4 7.7 6.5 4.8
257
Over – 0 mg/kg rat context CS1 CS2 CS3 CS4 CS mean 2 22.7 0 26.7 20 26.7 18.4 6 13.6 20 20 26.7 60 31.7 10 13.6 6.7 33.3 20 33.3 23.3 14 27.3 100 53.3 60 6.7 55.0 16 9.1 46.7 66.7 66.7 73.3 63.4 22 0 6.7 46.7 6.7 20 20.0 23 16.7 46.7 46.7 20 73.3 46.7 27 4.5 6.7 53.3 20 20 25.0 31 13.6 46.7 73.3 40 46.7 51.7 32 0 33.3 13.3 20 20 21.7 37 27.3 66.7 33.3 33.3 33.3 41.7 mean 13.5 34.6 42.430.3 37.6 36.2 sem 2.9 9.3 5.7 5.6 6.9 4.8
Over - 1 mg/kg rat context CS1 CS2 CS3 CS4 CS mean 3 0 13.3 20 6.7 6.7 11.7 8 45.5 46.7 46.7 33.3 73.3 50.0 11 18.2 13.3 26.7 6.7 73.3 30.0 15 27.3 6.7 40 33.3 26.7 26.7 18 18.2 20 13.3 0 0 8.3 25 4.5 0 6.7 0 0 1.7 34 0 26.7 13.3 26.7 13.3 20.0 39 31.8 66.7 93.3 93.3 66.7 80.0 mean 18.2 24.2 32.525.0 32.5 28.5 sem 5.8 7.9 10.0 11.0 11.7 9.1
Over - 10 mg/kg rat context CS1 CS2 CS3 CS4 CS mean 1.7 4.5 60 53.3 66.7 80 65.0 1.8 9.1 33.3 73.3 80 60 61.7 4.8 31.8 46.7 93.3 73.3 53.3 66.7 5.4 13.6 46.7 46.7 33.3 53.3 45.0 5.5 20 60 40 66.7 40 51.7 6.8 13.6 66.7 66.7 26.7 20 45.0 7.1 13.6 46.7 20 20 33.3 30.0 7.6 4.5 33.3 66.7 73.3 53.3 56.7 8.3 4.5 26.7 40 40 60 41.7 mean 12.8 46.7 55.653.3 50.4 51.5 sem 3.0 4.6 7.3 7.7 5.8 4.1
258 Statistical Analyses
Stage I: Comparison of mean freezing to context and mean CS freezing on Days 3 and 4
Number of Groups: 1 Number of Measurements: 4
Number of subjects in... Group 1: 40
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 1 1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 32.658 Measurement 1 2 3 4 Mean 14.778 8.855 58.330 48.670 SD 24.451 14.982 20.697 28.297 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 29265.647 39 750.401 ------Within ------W1 69501.401 1 69501.401 106.693 Error 25405.087 39 651.412 ------
259 Stage II: Comparison of context freezing and mean CS freezing
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 28
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 25.980 Measurement 1 2 Mean 5.839 46.121 SD 10.044 19.071 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 6253.466 27 231.610 ------Within ------W1 22717.114 1 22717.114 97.513 Error 6290.023 27 232.964 ------
260 Stage II: Test for summation
Number of Groups: 1 Number of Measurements: 3
Number of subjects in... Group 1: 28
Within contrast coefficients Contrast Measurement... 1 2 3 W1 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 54.638 Measurement 1 2 3 Mean 57.379 47.864 58.671 SD 21.196 24.954 20.717 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 23061.751 27 854.139 ------Within ------W1 683.247 1 683.247 2.314 Error 7973.700 27 295.322 ------
261 Stage II: CS freezing
Number of Groups: 1 Number of Measurements: 2
Number of subjects in... Group 1: 28
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 46.120 Measurement 1 2 Mean 58.671 33.568 SD 20.717 22.254 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 19638.933 27 727.368 ------Within ------W1 8822.650 1 8822.650 44.778 Error 5319.805 27 197.030 ------
262 Test: Comparison of context freezing to mean CS freezing
Number of Groups: 4 Number of Measurements: 5
Number of subjects in... Group 1 (Control): 12 Group 2 (Over - 0 mg/kg): 11 Group 3 (Over - 1 mg/kg): 8 Group 4 (Over - 10 mg/kg): 9
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 W1 4 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 42.277 Measurement 1 2 3 4 5 Mean 5.292 56.658 57.775 42.775 48.883 SD 8.642 20.397 25.471 26.587 22.442
Group 2 Overall Mean: 31.671 Measurement 1 2 3 4 5 Mean 13.491 34.564 42.418 30.309 37.573 SD 9.699 30.812 18.922 18.468 22.745
Group 3 Overall Mean: 26.473 Measurement 1 2 3 4 5 Mean 18.188 24.175 32.500 25.000 32.500 SD 16.315 22.251 28.151 30.996 33.122
263 Group 4 Overall Mean: 43.744 Measurement 1 2 3 4 5 Mean 12.800 46.678 55.556 53.333 50.356 SD 8.910 13.752 21.851 23.090 17.358
Means and SDs averaged across groups Measurement 1 2 3 4 5 Mean 12.443 40.519 47.062 37.854 42.328 SD 10.879 23.003 23.620 24.820 24.039 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 43052.556 36 1195.904 ------Within ------W1 27142.179 1 27142.179 138.742 Error 7042.718 36 195.631 ------
Test: Between-groups comparison of context freezing
Number of Groups: 4 Number of Measurements: 1
Number of subjects in... Group 1 (Control): 12 Group 2 (Over - 0 mg/kg): 11 Group 3 (Over - 1 mg/kg): 8 Group 4 (Over - 10/mg/kg): 9
264 Between contrast coefficients Contrast Group... 1 2 3 4 B1 3 -1 -1 -1 B2 0 2 -1 -1 B3 0 1 -1 0 B4 0 1 0 -1 B5 0 0 1 -1 B6 0 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 5.292 Measurement 1 Mean 5.292 SD 8.642
Group 2 Overall Mean: 13.491 Measurement 1 Mean 13.491 SD 9.699
Group 3 Overall Mean: 18.188 Measurement 1 Mean 18.188 SD 16.315
Group 4 Overall Mean: 12.800 Measurement 1 Mean 12.800 SD 8.910
265 Means and SDs averaged across groups Measurement 1 Mean 12.443 SD 10.879 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 759.647 1 759.647 6.419 B2 26.754 1 26.754 0.226 B3 102.163 1 102.163 0.863 B4 2.363 1 2.363 0.020 B5 122.930 1 122.930 1.039 B6 55.950 1 55.950 0.473 Error 4260.507 36 118.347 ------
Test: Between-group comparisons of CS freezing
Number of Groups: 4 Number of Measurements: 4
Number of subjects in... Group 1 (Control): 12 Group 2 (Over - 0 mg/kg): 11 Group 3 (Over - 1 mg/kg): 8 Group 4 (Over - 10 mg/kg): 9
266 Between contrast coefficients Contrast Group... 1 2 3 4 B1 3 -1 -1 -1 B2 0 1 -1 0 B3 0 1 1 -2
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 -3 -1 1 3
Means and Standard Deviations Group 1 Overall Mean: 51.523 Measurement 1 2 3 4 Mean 56.658 57.775 42.775 48.883 SD 20.397 25.471 26.587 22.442
Group 2 Overall Mean: 36.216 Measurement 1 2 3 4 Mean 34.564 42.418 30.309 37.573 SD 30.812 18.922 18.468 22.745
Group 3 Overall Mean: 28.544 Measurement 1 2 3 4 Mean 24.175 32.500 25.000 32.500 SD 22.251 28.151 30.996 33.122
Group 4 Overall Mean: 51.481 Measurement 1 2 3 4 Mean 46.678 55.556 53.333 50.356 SD 13.752 21.851 23.090 17.358
267 Means and SDs averaged across groups Measurement 1 2 3 4 Mean 40.519 47.062 37.854 42.328 SD 23.003 23.620 24.820 24.039 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 5456.078 1 5456.078 4.285 B2 1090.496 1 1090.496 0.857 B3 8839.816 1 8839.816 6.943 Error 45834.766 36 1273.188 ------Within ------W1 27.858 1 27.858 0.067 B1W1 886.405 1 886.405 2.124 B2W1 97.861 1 97.861 0.235 B3W1 0.789 1 0.789 0.002 Error 15020.620 36 417.239 ------
268 Test: Comparison of groups Over- 0 mg/kg and Over- 10 mg/kg
Group Statistics
VAR000 01 N Mean Std. Deviation Std. Error Mean VAR00002 1.00 11 36.2159 16.04089 4.83651 2.00 9 51.4806 12.16707 4.05569
Independent Samples Test
Levene's Test for Equality of Variances Sig. (2- F Sig. t df tailed)
Lower Upper Lower Upper Lower VAR0000 Equal 2 variances 2.215 .154 -2.351 18 .030 assumed Equal variances not -2.418 17.928 .026 assumed
269 Appendix 4: Experiment 4 Raw Data Stage I Conditioning Tone Control - veh Day 1 Day 3 rat context CS context CS 2 4.5 0 36.4 53.3 4 0 0 18.2 66.7 13 0 33.3 4.5 86.7 14 0 6.7 4.5 80 22 0 35.7 9.1 80.0 24 9.1 14.3 45.5 46.7 29 0 0 22.7 85.7 31 0 26.7 9.1 50.0 32 9.1 0 31.8 66.7 34 0 0 0.0 86.7 mean 2.3 11.7 18.2 70.3 sem 1.2 4.7 4.9 5.0
Control - FG Day 1 Day 3 rat context CS context CS 1 4.5 53.3 9.1 66.7 3 0 0 36.4 66.7 12 4.5 6.7 0 73.3 19 0 40 4.5 46.7 23 0 21.4 0.0 60.0 28 0 0 4.5 78.6 30 0 0 4.5 42.9 33 0 26.7 54.5 86.7 39 4.5 0.0 9.1 93.3 40 0.0 0.0 27.3 60.0 mean 1.4 14.8 15.0 67.5 sem 0.7 6.2 5.8 5.1
Over - Veh Day 1 Day 3 rat context CS context CS 6 0 0 0 73.3 8 0 0 40.9 80 9 0 13.3 31.8 73.3 16 0 0 9.1 60 17 0 0 0 53.3 25 0 42.9 0.0 33.3 26 0 7.1 9.1 73.3 35 4.5 60.0 13.6 86.7 36 0.0 0.0 0.0 93.3 38 0.0 0.0 36.4 46.7 mean 0.5 12.3 14.1 67.3 sem 0.5 6.8 5.1 5.9
270 Over - FG Day 1 Day 3 rat context CS context CS 5 0 0 18.2 60 7 0 0 4.5 73.3 10 0 0 4.5 66.7 11 0 0 0 53.3 15 0 66.7 18.2 66.7 18 0 6.7 22.7 93.3 20 0 0 18.2 73.3 21 0 40 31.8 86.7 27 0 57.1 31.8 60.0 37 4.5 0.0 4.5 53.3 mean 0.5 17.1 15.4 68.7 sem 0.5 8.5 3.7 4.2
Flashing Light Control -Veh Day 2 Day 4 rat context CS context CS 2 9.1 53.3 27.3 93.3 4 4.5 0 13.6 40 13 0 73.3 0 73.3 14 9.1 0 0 40 22 4.5 60.0 0.0 86.7 24 0.0 93.3 0.0 46.7 29 18.2 0.0 18.2 80.0 31 18.2 0.0 9.1 86.7 32 18.2 0.0 33.3 60.0 34 0.0 0.0 26.7 6.7 mean 8.2 28.0 12.8 61.3 sem 2.1 10.4 3.6 7.7
Control - FG Day 2 Day 4 rat context CS context CS 1 31.8 13.3 36.4 60 3 12.6 0 18.2 26.7 12 0 66.7 0 80 19 4.5 86.7 13.6 73.3 23 0.0 0.0 0.0 60.0 28 0.0 0.0 0.0 66.7 30 0.0 13.3 13.6 13.3 33 31.8 13.3 3.3 60.0 39 4.5 6.7 26.7 53.3 40 14 20.0 26.7 66.7 mean 9.9 22.0 13.9 56.0 sem 3.4 8.0 3.5 5.5
271 Over - Veh Day 2 Day 4 rat context CS context CS 6 0 33.3 0 80 8 9.1 0 0 13.3 9 22.7 6.7 3.3 53.3 16 18.2 0 27.3 53.3 17 4.5 26.7 0 80 25 13.6 20.0 6.7 66.7 26 22.7 0.0 0.0 13.3 35 31.8 92.3 0.0 86.7 36 0.0 7.7 0.0 53.3 38 13.6 20.0 9.1 80.0 mean 13.6 20.7 4.6 58.0 sem 2.9 7.7 2.4 7.4
Over - FG Day 2 Day 4 rat context CS context CS 5 4.5 26.7 0 73.3 7 0 0 9.1 53.3 10 9.1 20 10 86.7 11 0 0 0 33.3 15 18.2 66.7 0 53.3 18 4.5 93.3 13.6 66.7 20 0.0 6.7 6.7 40.0 21 31.8 33.3 6.7 66.7 27 27.3 0.0 13.3 73.3 37 0 15.4 0.0 66.7 mean 9.5 26.2 5.9 61.3 sem 3.3 8.7 1.6 4.5
Stage II conditioning Over - Veh Over - FG rat context CS1 CS2 rat context CS1 CS2 6 0 73.3 53.3 5 31.8 60 40 8 9.1 80 80 7 0 73.3 33.3 9 20 33.3 60 10 6.7 86.7 53.3 16 9.1 66.7 26.7 11 0 73.3 46.7 17 4.5 53.3 6.7 15 9.1 46.7 13.3 25 20 53.3 53.3 18 31.8 80 26.7 26 0 66.7 26.7 20 0.0 73.3 100.0 35 4.5 100 80 21 50.0 93.3 53.3 36 0 85.7 66.7 27 0.0 86.7 26.7 38 0 85.7 73.3 37 0.0 42.9 73.3 mean 6.7 69.8 52.7 mean 12.9 71.6 46.7 sem 2.2 5.4 6.9 sem 5.0 4.7 7.0
272 Test Control - Veh rat context CS1 CS2 CS3 CS4 CS mean 2 18.2 33.3 46.7 46.7 53.3 45.0 4 4.5 86.7 73.3 26.7 33.3 55.0 13 0 73.3 100 33.3 66.7 68.3 14 0 60 66.7 80 66.7 68.4 22 0 66.7 73.3 40 80 65.0 24 0 60 86.7 46.7 0 48.4 29 0 86.7 93.3 60 73.3 78.3 31 0 60 53.3 66.7 80 65.0 32 10 86.7 86.7 86.7 66.7 81.7 34 3.3 46.7 46.7 80 46.7 55.0 mean 3.6 66.072.7 56.7 56.7 63.0 sem 1.9 5.76.1 6.7 7.8 3.8
Control - FG rat context CS1 CS2 CS3 CS4 CS mean 1 9.1 33.3 53.3 46.7 26.7 40.0 3 0 26.7 53.3 40 40 40.0 12 0 40 80 53.3 60 58.3 19 27.3 60 86.7 73.3 33.3 63.3 23 0 66.7 66.7 53.3 73.3 65.0 28 0 80 80 40 26.7 56.7 30 13.6 33.3 60 46.7 40 45.0 33 40 100 93.3 93.3 93.3 95.0 39 0 73.3 66.7 80 66.7 71.7 40 0 66.7 80 40 46.7 58.4 mean 9.0 58.072.0 56.7 50.7 59.3 sem 4.5 7.64.4 6.0 7.0 5.2
Over - Veh rat context CS1 CS2 CS3 CS4 CS mean 6 18.2 26.7 6.7 20 20 18.4 8 4.5 26.7 53.3 66.7 40 46.7 9 36.7 60 46.7 60 53.3 55.0 16 13.6 66.7 26.7 13.3 0 26.7 17 4.5 33.3 33.3 26.7 26.7 30.0 25 10 46.7 66.7 40 46.7 50.0 26 0 46.7 53.3 20 6.7 31.7 35 0 66.7 6.7 6.7 13.3 23.4 36 0 46.7 60 40 46.7 48.4 38 13.6 86.7 66.7 33.3 46.7 58.4 mean 10.1 50.742.0 32.7 30.0 38.8 sem 3.6 6.17.2 6.2 6.1 4.5
273 Over - FG rat context CS1 CS2 CS3 CS4 CS mean 5 4.5 40 33.3 53.3 26.7 38.3 7 13.6 53.3 66.7 26.7 26.7 43.4 10 40 93.3 93.3 86.7 86.7 90.0 11 0 40 40 33.3 40 38.3 15 40.9 53.3 33.3 6.7 0 23.3 18 13.6 53.3 86.7 73.3 66.7 70.0 20 3.3 73.3 46.7 53.3 46.7 55.0 21 23.3 53.3 46.7 60 26.7 46.7 27 3.3 33.3 80 80 66.7 65.0 37 0 33.3 60 86.7 53.3 58.3 mean 14.3 52.6 58.756.0 44.0 52.8 sem 4.9 5.9 7.08.6 8.0 6.0
Statistical Analyses
Stage I: Comparison of mean context freezing and mean CS freezing on days 3 and 4
Number of Groups: 1 Number of Measurements: 4
Number of subjects in... Group 1: 40
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 1 -1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 38.146 Measurement 1 2 3 4 Mean 15.675 68.430 9.313 59.165 SD 15.064 15.526 10.952 22.510 ------
274 Analysis of Variance Summary Table Source SS df MS F ------Between 10177.034 39 260.950 ------Within ------W1 105282.991 1 105282.991 371.310 Error 11058.247 39 283.545 ------
Stage II: Comparison of context freezing and mean CS freezing
Number of Groups: 2 Number of Measurements: 3
Number of subjects in... Group 1: 10 Group 2: 10
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 W1 2 -1 -1
275 Means and Standard Deviations Group 1 Overall Mean: 43.063 Measurement 1 2 3 Mean 6.720 69.800 52.670 SD 7.850 19.522 25.018
Group 2 Overall Mean: 43.740 Measurement 1 2 3 Mean 12.940 71.620 46.660 SD 18.183 16.946 25.332
Means and SDs averaged across groups Measurement 1 2 3 Mean 9.830 70.710 49.665 SD 14.005 18.279 25.176 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 6.868 1 6.868 0.015 Error 7998.548 18 444.364 ------Within ------W1 33811.704 1 33811.704 93.219 B1W1 230.464 1 230.464 0.635 Error 6528.830 18 362.713 ------
276 Stage II: Test for summation
Number of Groups: 2 Number of Measurements: 3 Number of subjects in... Group 1 (Over – Veh): 10 Group 2 (Over – FG): 10
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 W1 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 65.037 Measurement 1 2 3 Mean 67.320 57.990 69.800 SD 18.712 26.695 19.522
Group 2 Overall Mean: 67.203 Measurement 1 2 3 Mean 68.660 61.330 71.620 SD 13.355 16.283 16.946
Means and SDs averaged across groups Measurement 1 2 3 Mean 67.990 59.660 70.710 SD 16.256 22.111 18.279 ------
277 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 70.417 1 70.417 0.156 Error 8132.013 18 451.778 ------Within ------W1 632.043 1 632.043 3.214 B1W1 0.901 1 0.901 0.005 Error 3540.122 18 196.673 ------
Stage II: CS freezing
Number of Groups: 2 Number of Measurements: 2 Number of subjects in... Group 1 (Over - Veh): 10 Group 2 (Over - FG): 10
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
278 Means and Standard Deviations Group 1 Overall Mean: 61.235 Measurement 1 2 Mean 69.800 52.670 SD 19.522 25.018
Group 2 Overall Mean: 59.140 Measurement 1 2 Mean 71.620 46.660 SD 16.946 25.332
Means and SDs averaged across groups Measurement 1 2 Mean 70.710 49.665 SD 18.279 25.176 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 43.890 1 43.890 0.072 Error 10997.099 18 610.950 ------Within ------W1 4428.920 1 4428.920 12.406 B1W1 153.272 1 153.272 0.429 Error 6425.723 18 356.985 ------
279 Test: Between-groups comparison of context freezing
Number of Groups: 4 Number of Measurements: 1
Number of subjects in... Group 1 (Control - Veh): 10 Group 2 (Control - FG): 10 Group 3 (Over - Veh): 10 Group 4 (Over - FG): 10
Between contrast coefficients Contrast Group... 1 2 3 4 B1 1 1 -1 -1 B2 1 -1 1 -1 B3 1 -1 -1 1
Means and Standard Deviations Group 1 Overall Mean: 3.600 Measurement 1 Mean 3.600 SD 6.081
Group 2 Overall Mean: 9.000 Measurement 1 Mean 9.000 SD 14.154
Group 3 Overall Mean: 10.110 Measurement 1 Mean 10.110 SD 11.384
280 Group 4 Overall Mean: 14.250 Measurement 1 Mean 14.250 SD 15.624
Means and SDs averaged across groups Measurement 1 Mean 9.240 SD 12.360 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 345.744 1 345.744 2.263 B2 227.529 1 227.529 1.489 B3 3.969 1 3.969 0.026 Error 5499.294 36 152.758 ------
Test: Comparison of context freezing to mean CS freezing
Number of Groups: 4 Number of Measurements: 2
Number of subjects in... Group 1 (Control - Veh): 10 Group 2 (Control - FG): 10 Group 3 (Over - Veh): 10 Group 4 (Over - FG): 10
281 Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 33.250 Measurement 1 2 Mean 3.600 62.900 SD 6.081 12.115
Group 2 Overall Mean: 34.150 Measurement 1 2 Mean 9.000 59.300 SD 14.154 16.466
Group 3 Overall Mean: 24.455 Measurement 1 2 Mean 10.110 38.800 SD 11.384 14.351
Group 4 Overall Mean: 33.475 Measurement 1 2 Mean 14.250 52.700 SD 15.624 19.195
Means and SDs averaged across groups Measurement 1 2 Mean 9.240 53.425 SD 12.360 15.750 ------
282 Analysis of Variance Summary Table Source SS df MS F ------Between 8833.267 36 245.369 ------Within ------W1 39046.285 1 39046.285 251.159 Error 5596.727 36 155.465 ------
Test: CS freezing
Number of Groups: 4 Number of Measurements: 4
Number of subjects in... Group 1 (Control - Veh): 10 Group 2 (Control - FG): 10 Group 3 (Over - Veh): 10 Group 4 (Over - FG): 10
Between contrast coefficients Contrast Group... 1 2 3 4 B1 1 -1 0 0 B2 1 1 -2 0 B3 1 1 0 -2
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 -3 -1 1 3
283 Means and Standard Deviations Group 1 Overall Mean: 63.008 Measurement 1 2 3 4 Mean 66.010 72.670 56.680 56.670 SD 17.917 19.226 21.143 24.800
Group 2 Overall Mean: 59.333 Measurement 1 2 3 4 Mean 58.000 72.000 56.660 50.670 SD 23.947 13.989 18.909 22.030
Group 3 Overall Mean: 38.845 Measurement 1 2 3 4 Mean 50.690 42.010 32.670 30.010 SD 19.429 22.666 19.491 19.184
Group 4 Overall Mean: 52.833 Measurement 1 2 3 4 Mean 52.640 58.670 56.000 44.020 SD 18.707 22.182 27.070 25.380
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 56.835 61.338 50.503 45.343 SD 20.136 19.819 21.893 22.981 ------
284 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 270.113 1 270.113 0.274 B2 13290.817 1 13290.817 13.466 B3 1853.704 1 1853.704 1.878 Error 35531.895 36 986.997 ------Within ------W1 4106.445 1 4106.445 10.209 B1W1 11.156 1 11.156 0.028 B2W1 314.368 1 314.368 0.782 B3W1 49.127 1 49.127 0.122 Error 14480.523 36 402.237 ------
Experiment 4: Comparison of groups Over- Veh and Over – FG
Group Statistics
VAR0000 Std. Std. Error 1 N Mean Deviation Mean VAR0000 1.00 10 38.8450 14.36031 4.54113 2 2.00 10 52.8325 19.08612 6.03556
285 Independent Samples Test
Levene's Test for Equality of Variances Sig. (2- F Sig. t df tailed)
Lower Upper Lower Upper Lower VAR00002 Equal variances .279 .604 -1.852 18 .081 assumed Equal variances not -1.852 16.717 .082 assumed
286 Appendix 5: Experiment 5 Raw Data Conditioning Low Shock - Veh rat context CS1 CS2 1 0 0 0 4 4.5 0 20 12 9.1 6.7 100 14 0 0 53.3 16 0 0 80 19 0 6.7 73.3 21 0 0 60 25 0 0 80 26 0 6.7 86.7 30 3.3 6.7 80 mean 1.7 2.7 63.3 sem 1.0 1.1 9.9
Low Shock - FG rat context CS1 CS2 2 0 6.7 66.7 6 0 0 100 10 13.6 0 73.3 13 0 13.3 86.7 15 0 0 80 17 0 0 26.7 18 6.7 0 60 23 0 0 0 24 0 0 93.3 27 0 13.3 80 mean 2.0 3.3 66.7 sem 1.4 1.8 9.8
High Shock rat context CS1 CS2 3 0 0 66.7 5 0 0 93.3 7 0 13.3 73.3 8 0 0 53.3 9 0 0 46.7 11 0 13.3 80 20 0 26.7 93.3 22 0 6.7 80 28 0 6.7 66.7 29 0 6.7 93.3 mean 0.0 7.3 74.7 sem 0.0 2.7 5.2
287 Test Low Shock - Veh rat context CS1 CS2 CS3 CS4 CS5 CS6 CS mean 1 4.5 66.7 80 80 86.7 66.7 0 63.4 4 0 26.7 6.7 20 0 0 13.3 11.1 12 0 40 46.7 20 6.7 0 0 18.9 14 22.7 100 100 100 80 26.7 66.7 78.9 16 0 60 26.7 46.7 93.3 53.3 26.7 51.1 19 0 53.3 80 73.3 20 26.7 0 42.2 21 0 20 20 0 26.7 0 53.3 20.0 25 4.5 73.3 80 93.3 60 80 6.7 65.6 26 0 86.7 93.3 100 93.3 100 73.3 91.1 30 9.1 93.3 66.7 100 86.7 86.7 93.3 87.8 mean 4.1 62.0 60.0 63.3 55.3 44.0 33.3 53.0 sem 2.3 8.7 10.4 12.2 12.0 12.1 11.2 9.3
Low Shock - FG rat context CS1 CS2 CS3 CS4 CS5 CS6 CS mean 2 4.5 73.3 20 66.7 13.3 40 13.3 37.8 6 9.1 100 86.7 86.7 60 73.3 46.7 75.6 10 0 46.7 80 73.3 33.3 86.7 33.3 58.9 13 0 80 13.3 0 0 13.3 0 17.8 15 4.5 13.3 20 20 0 26.7 20 16.7 17 0 60 66.7 33.3 80 20 6.7 44.5 18 0 46.7 66.70000 18.9 23 0 33.3 33.3 13.3 0 0 0 13.3 24 0 60 80 86.7 86.7 80 73.3 77.8 27 0 73.3 66.7 66.7 73.3 80 26.7 64.5 mean 1.8 58.7 53.3 44.7 34.7 42.0 22.0 42.6 sem 1.0 7.9 9.0 11.1 11.6 11.0 7.6 8.0
High Shock rat context CS1 CS2 CS3 CS4 CS5 CS6 CS mean 3 9.1 40 80 73.3 73.3 66.7 86.7 70.0 5 13.6 46.7 60 60 66.7 80 53.3 61.1 7 36.4 73.3 93.3 100 100 93.3 100 93.3 8 45.5 66.7 86.7 80 66.7 53.3 20 62.2 9 9.1 46.7 66.7 60 53.3 66.7 33.3 54.5 11 18.2 86.7 86.7 80 86.7 73.3 86.7 83.4 20 31.8 93.3 100 73.3 80 86.7 33.3 77.8 22 18.2 86.7 93.3 100 60 93.3 73.3 84.4 28 18.2 86.7 93.3 100 93.3 100 93.3 94.4 29 40.9 93.3 53.3 80 86.7 93.3 93.3 83.3 mean 24.1 72.0 81.3 80.7 76.7 80.7 67.3 76.4 sem 4.2 6.6 5.0 4.8 4.8 4.8 9.4 4.4
288 Statistical Analyses
Test: Comparison of context freezing to mean CS freezing
Number of Groups: 3 Number of Measurements: 7
Number of subjects in... Group 1 (Low Shock - Veh): 10 Group 2 (Low Shock - FG): 10 Group 3 (High Shock): 10
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 6 7 W1 6 -1 -1 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 46.014 Measurement 1 2 3 4 5 6 7 Mean 4.080 62.000 60.010 63.330 55.340 44.010 33.330 SD 7.243 27.401 32.796 38.517 37.971 38.389 35.266
Group 2 Overall Mean: 36.734 Measurement 1 2 3 4 5 6 7 Mean 1.810 58.660 53.340 44.670 34.660 42.000 22.000 SD 3.172 24.907 28.476 35.018 36.761 34.864 23.938
Group 3 Overall Mean: 68.964 Measurement 1 2 3 4 5 6 7 Mean 24.100 72.010 81.330 80.660 76.670 80.660 67.320 SD 13.398 20.789 15.960 15.222 15.156 15.212 29.731
289 Means and SDs averaged across groups Measurement 1 2 3 4 5 6 7 Mean 9.997 64.223 64.893 62.887 55.557 55.557 40.883 SD 8.982 24.518 26.716 31.313 31.743 31.202 30.004 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 81501.018 27 3018.556 ------Within ------W1 57619.543 1 57619.543 127.372 Error 12214.040 27 452.372 ------
Test: Between-groups comparison of context freezing
Number of Groups: 3 Number of Measurements: 1
Number of subjects in... Group 1 (Low Shock - Veh): 10 Group 2 (Low Shock - FG): 10 Group 3 (High Shock): 10
Between contrast coefficients Contrast Group... 1 2 3 B1 1 1 -2 B2 1 -1 0
290 Means and Standard Deviations Group 1 Overall Mean: 4.080 Measurement 1 Mean 4.080 SD 7.243
Group 2 Overall Mean: 1.810 Measurement 1 Mean 1.810 SD 3.172
Group 3 Overall Mean: 24.100 Measurement 1 Mean 24.100 SD 13.398
Means and SDs averaged across groups Measurement 1 Mean 9.997 SD 8.982 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2983.560 1 2983.560 36.984 B2 25.765 1 25.765 0.319 Error 2178.145 27 80.672 ------
291 Test: CS freezing
Number of Groups: 3 Number of Measurements: 6
Number of subjects in... Group 1 (Low Shock - Veh): 10 Group 2 (Low Shock - FG): 10 Group 3 (High Shock): 10
Between contrast coefficients Contrast Group... 1 2 3 B1 1 1 -2 B2 1 -1 0
Means and Standard Deviations Group 1 Overall Mean: 53.003 Measurement 1 2 3 4 5 6 Mean 62.000 60.010 63.330 55.340 44.010 33.330 SD 27.401 32.796 38.517 37.971 38.389 35.266
Group 2 Overall Mean: 42.555 Measurement 1 2 3 4 5 6 Mean 58.660 53.340 44.670 34.660 42.000 22.000 SD 24.907 28.476 35.018 36.761 34.864 23.938
Group 3 Overall Mean: 76.442 Measurement 1 2 3 4 5 6 Mean 72.010 81.330 80.660 76.670 80.660 67.320 SD 20.789 15.960 15.222 15.156 15.212 29.731
292 Means and SDs averaged across groups Measurement 1 2 3 4 5 6 Mean 64.223 64.893 62.887 55.557 55.557 40.883 SD 24.518 26.716 31.313 31.743 31.202 30.004 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 32861.556 1 32861.556 9.693 B2 3275.030 1 3275.030 0.966 Error 91536.914 27 3390.256 ------
293 Appendix 6: Experiment 6a Raw Data Stage I Conditioning Day 1 rat pre-CS CS1 CS2 CS3 CS4 3 5.3 26.7 86.7 73.3 33.3 6 1.3 0 93.3 53.3 93.3 7 2.7 0 86.7 93.3 93.3 8 0 0 80 80 60 9 0 13.3 41.7 86.7 7.7 10 5.3 0 83.3 86.7 76.9 12 0 0 75 93.3 38.5 15 0 0 41.7 93.3 84.6 mean 1.8 5.0 73.682.5 61.0 sem 0.8 3.5 7.2 4.9 11.2
Day 2 rat pre-CS CS1 CS2 CS3 CS4 3 20 53.3 53.3 54.5 30.8 6 22.7 80 60 63.6 76.9 7 74.7 86.7 100 18.2 76.9 8 22.7 73.3 66.7 27.3 46.2 9 21.3 93.3 92.9 30.8 38.5 10 12 100 78.6 53.8 53.8 12 33.3 80 50 38.5 23.1 15 6.7 60 57.1 15.4 23.1 mean 26.7 78.3 69.8 37.8 46.2 sem 7.4 5.6 6.6 6.3 7.7
Day 3 rat pre-CS CS1 CS2 CS3 CS4 3 37.3 46.2 57.1 28.6 35.7 6 22.7 75 42.9 30.8 46.2 7 74.7 91.7 69.2 61.5 46.2 8 14.7 58.3 46.2 61.5 38.5 9 14.7 41.7 16.7 0 50 10 20 8.3 58.3 50 33.3 12 44 72.7 54.5 33.3 58.3 15 12 18.2 72.7 58.3 33.3 mean 30.0 51.5 52.2 40.5 42.7 sem 7.6 10.1 6.27.6 3.2
294 Stage II conditioning Blocking Control rat pre-CS CS1 CS2 rat pre-CS CS1 CS2 3 33.3 13.3 13.3 1 0 0 80 6 36 13.3 0 2 0 0 73.3 7 58.7 0 0 4 0 13.3 13.3 8 4 4053.3 5 0 0 73.3 9 28 6.7 33.3 11 0 6.7 60 10 28 40 26.7 13 1.3 0 86.7 12 45.3 60 40 14 0 33.3 80 15 13.3 13.3 46.7 16 0 0 73.3 mean 30.8 23.326.7 mean 0.2 6.7 67.5 sem 6.1 7.3 7.2 sem 0.2 4.2 8.2
Test Blocking rat pre-CS CS1 CS2 CS3 CS4 CS mean 3 29.3 0 6.7 13.3 13.3 8.3 6 30.7 33.3 40 26.7 13.3 28.3 7 26.7 40 40 33.3 26.7 35.0 8 18.7 20 20 20 6.7 16.7 9 14.7 13.3 6.7 0 0 5.0 10 0 6.7 0 0 0 1.7 12 28 40 40 20 0 25.0 15 12 20 26.7 13.3 0 15.0 mean 20.0 21.7 22.5 15.8 7.5 16.9 sem 3.8 5.3 5.9 4.2 3.4 4.2
Control rat pre-CS CS1 CS2 CS3 CS4 CS mean 1 2.7 46.7 20 60 46.7 43.4 2 8 66.7 86.7 93.3 93.3 85.0 4 10.7 26.7 60 33.3 33.3 38.3 5 2.7 46.7 80 73.3 66.7 66.7 11 4 26.7 53.3 33.3 13.3 31.7 13 0 13.3 20 20 60 28.3 14 9.3 6.7 46.7 46.7 40 35.0 16 0 33.3 33.3 60 40 41.7 mean 4.7 33.4 50.0 52.5 49.2 46.3 sem 1.5 6.9 8.9 8.5 8.5 6.9
295 Statistical Analyses
Test: Comparison of context freezing to mean CS freezing
Number of Groups: 2 Number of Measurements: 5
Number of subjects in... Group 1 (Blocking): 8 Group 2 (Control): 8
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 W1 4 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 17.503 Measurement 1 2 3 4 5 Mean 20.013 21.663 22.513 15.825 7.500 SD 10.715 15.007 16.683 11.784 9.722
Group 2 Overall Mean: 37.935 Measurement 1 2 3 4 5 Mean 4.675 33.350 50.000 52.488 49.163 SD 4.153 19.528 25.206 24.015 24.157
Means and SDs averaged across groups Measurement 1 2 3 4 5 Mean 12.344 27.506 36.256 34.156 28.331 SD 8.126 17.415 21.374 18.915 18.413 ------
296 Analysis of Variance Summary Table Source SS df MS F ------Between 13006.037 14 929.003 ------Within ------W1 4727.813 1 4727.813 26.454 Error 2502.057 14 178.718 ------
Test: Between-groups comparison of context freezing
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Blocking): 8 Group 2 (Control): 8
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 20.013 Measurement 1 Mean 20.013 SD 10.715
297 Group 2 Overall Mean: 4.675 Measurement 1 Mean 4.675 SD 4.153
Means and SDs averaged across groups Measurement 1 Mean 12.344 SD 8.126 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 940.956 1 940.956 14.251 Error 924.364 14 66.026 ------
Test: CS freezing
Number of Groups: 2 Number of Measurements: 4
Number of subjects in... Group 1 (Blocking): 8 Group 2 (Control): 8
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
298 Means and Standard Deviations Group 1 Overall Mean: 16.875 Measurement 1 2 3 4 Mean 21.663 22.513 15.825 7.500 SD 15.007 16.683 11.784 9.722
Group 2 Overall Mean: 46.250 Measurement 1 2 3 4 Mean 33.350 50.000 52.488 49.163 SD 19.528 25.206 24.015 24.157
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 27.506 36.256 34.156 28.331 SD 17.415 21.374 18.915 18.413 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 13806.250 1 13806.250 13.254 Error 14583.730 14 1041.695 ------
299 Appendix 7: Experiment 6b Raw Data Stage I conditioning Day 1 Block – 0 mg/kg rat context CS1 CS2 CS3 CS4 3 0 0 86.7 86.7 73.3 8 0 13.3 6.7 73.3 66.7 11 36.4 0 93.3 100 53.3 14 0 0 86.7 100 100 17 0 0 93.3 20 86.7 28 0 0 6.7 60 40 30 0 0 100 93.3 80 mean 5.2 1.9 67.6 76.2 71.4 sem 5.2 1.9 15.8 10.9 7.7
Block - 1 mg/kg Rat context CS1 CS2 CS3 CS4 5 0 0 0 26.7 53.3 6 45.5 0 93.3 100 100 9 0 0 80 100 53.3 16 0 0 93.3 100 73.3 18 0 13.3 86.7 53.3 46.7 22 13.6 26.7 53.3 93.3 20 27 0 6.7 100 73.3 60 29 4.5 0 60 66.7 53.3 Mean 8.0 5.8 70.8 76.7 57.5 Sem 5.6 3.4 11.7 9.5 8.1
Block - 10 mg/kg Rat context CS1 CS2 CS3 CS4 1 4.5 13.3 80 73.3 78.6 2 0 0 46.7 66.7 66.7 12 0 20 33.3 66.7 66.7 13 0 6.7 93.3 100 73.3 19 4.5 13.3 86.7 60 60 20 0 6.7 60 100 93.3 24 0 0 53.3 26.7 46.7 25 0 0 93.3 53.3 73.3 Mean 1.1 7.5 68.3 68.3 69.8 Sem 0.7 2.7 8.1 8.5 4.8
300 Day 2 Block – 0 mg/kg Rat context CS1 CS2 CS3 CS4 3 4.5 33.3 66.7 46.7 33.3 8 0 73.3 46.7 13.3 46.7 11 0 46.7 26.7 53.3 13.3 14 0 80 100 80 86.7 17 4.5 71.4 46.7 46.7 13.3 28 0.0 21.4 20.0 7.1 20.0 30 0.0 42.9 66.7 50.0 46.7 Mean 1.3 52.7 53.4 42.4 37.1 Sem 0.8 8.5 10.3 9.4 9.9
Block - 1 mg/kg Rat context CS1 CS2 CS3 CS4 5 0 20 46.7 46.7 26.7 6 18.2 86.7 53.3 80 66.7 9 4.5 66.7 6.7 20 40 16 40.9 86.7 80 53.3 0 18 4.5 28.6 33.3 0.0 13.3 22 0.0 66.7 40.0 20.0 26.7 27 0.0 60.0 20.0 40.0 13.3 29 0.0 28.6 0.0 64.3 53.3 Mean 8.5 55.5 35.0 40.5 30.0 Sem 5.1 9.4 9.2 9.3 7.9
Block - 10 mg/kg Rat context CS1 CS2 CS3 CS4 1 9.1 73.3 53.3 26.7 33.3 2 0 26.7 40 40 40 12 4.5 20 66.7 66.7 66.7 13 13.6 86.7 40 60 53.3 19 18.2 35.7 46.7 13.3 40.0 20 0.0 93.3 53.3 46.7 26.7 24 4.5 42.9 40.0 35.7 26.7 25 13.6 46.7 13.3 40.0 40.0 Mean 7.9 53.2 44.2 41.1 40.8 Sem 2.4 9.8 5.5 6.1 4.8
301 Day 3 Block – 0 mg/kg Rat context CS1 CS2 CS3 CS4 3 4.5 20 20 6.7 40 8 4.5 26.7 6.7 20 6.7 11 4.5 53.3 6.7 0 6.7 14 36.4 86.7 40 53.3 33.3 17 9.1 40.0 13.3 6.7 13.3 28 9.1 33.3 33.3 60.0 6.7 30 0.0 53.3 40.0 33.3 33.3 Mean 9.7 44.8 22.9 25.7 20.0 Sem 4.6 8.4 5.6 9.0 5.6
Block - 1 mg/kg Rat context CS1 CS2 CS3 CS4 5 9.1 20 20 46.7 6.7 6 31.8 60 40 66.7 86.7 9 9.1 33.3 13.3 13.3 0 16 22.7 60 26.7 20 6.7 18 4.5 20.0 26.7 33.3 13.3 22 4.5 53.3 13.3 0.0 6.7 27 0.0 20.0 33.3 6.7 26.7 29 4.5 60.0 40.0 53.3 26.7 Mean 10.8 40.8 26.7 30.0 21.7 Sem 3.8 6.8 3.8 8.5 9.9
Block - 10 mg/kg Rat context CS1 CS2 CS3 CS4 1 9.1 33.3 33.3 26.7 6.7 2 13.6 60 20 20 0 12 18.2 6.7 6.7 0 26.7 13 18.2 93.3 53.3 46.7 20 19 18.2 60.0 20.0 13.3 13.3 20 9.1 46.7 33.3 53.3 40.0 24 4.5 40.0 53.3 26.7 40.0 25 13.6 60.0 26.7 20.0 26.7 Mean 13.1 50.0 30.8 25.8 21.7 Sem 1.8 8.9 5.8 6.1 5.2
302 Stage II conditioning Block – Control 0 mg/kg rat context CS1 CS2 rat context CS1 CS2 4 0 21.4 86.7 3 0 35.7 13.3 7 0 0 73.3 8 4.5 33.3 6.7 10 0 6.7 93.3 11 9.1 86.7 100 15 0 0 93.3 14 27.3 80 33.3 21 0 0 33.3 17 9.1 46.7 33.3 23 0 20 93.3 28 0.0 26.7 33.3 26 4.5 6.7 80 30 0.0 66.7 53.3 mean 0.6 7.8 79.0 mean 7.1 53.739.0 sem 0.6 3.5 8.2 sem 3.7 9.111.7
Block – Block – 1 mg/kg 10 mg/kg rat context CS1 CS2 rat context CS1 CS2 5 4.5 33.3 20 1 27.3 57.1 26.7 6 36.4 80 60 2 0 50 46.7 9 4.5 40 20 12 22.7 40 40 16 13.6 46.7 60 13 36.4 46.7 46.7 18 4.5 33.3 20.0 19 18.2 20.0 20.0 22 13.6 46.7 33.3 20 0.0 73.3 40.0 27 0.0 13.3 26.7 24 4.5 46.7 46.7 29 0.0 80.0 26.7 25 4.5 46.7 60.0 mean 9.6 46.7 33.3 mean 14.2 47.640.9 sem 4.3 8.2 6.0 sem 4.9 5.3 4.4
Test Control rat context CS1 CS2 CS3 CS4 CS mean 4 22.7 73.3 100 26.7 53.3 63.3 7 9.1 66.7 100 60 60 71.7 10 0 46.7 53.3 66.7 40 51.7 15 27.2 86.7 73.3 86.7 86.7 83.4 21 4.5 13.3 33.3 46.7 73.3 41.7 23 0 100 100 100 100 100.0 26 9.1 73.3 73.3 86.7 80 78.3 mean 10.4 65.7 76.2 67.6 70.5 70.0 sem 4.0 10.7 9.9 9.7 7.8 7.5
303 Block – 0 mg/kg rat context CS1 CS2 CS3 CS4 CS mean 3 9.1 46.7 13.3 20 20 25.0 8 4.5 33.3 0 13.3 6.7 13.3 11 22.7 33.3 73.3 33.3 0 35.0 14 36.4 60.0 66.7 100.0 66.7 73.4 17 0.0 13.3 66.7 0.0 40.0 30.0 28 27.3 13.3 0.0 6.7 6.7 6.7 30 9.1 73.3 86.7 53.3 26.7 60.0 mean 15.6 39.0 43.832.4 23.8 34.8 sem 5.0 8.5 14.213.1 8.8 9.1
Block - 1 mg/kg Rat context CS1 CS2 CS3 CS4 CS mean 5 0 40 46.7 20 13.3 30.0 6 31.8 66.7 73.3 80 53.3 68.3 9 4.5 26.7 53.3 13.3 33.3 31.7 16 27.3 66.7 86.7 73.3 66.7 73.4 18 4.5 26.7 60 66.7 53.3 51.7 22 4.5 6.7 26.7 33.3 46.7 28.4 27 4.5 60 46.7 53.3 53.3 53.3 29 13.6 33.3 40 6.7 40 30.0 Mean 11.3 40.9 54.243.3 45.0 45.8 Sem 4.2 7.7 6.710.2 5.7 6.5
Block - 10 mg/kg rat context CS1 CS2 CS3 CS4 CS mean 1 4.5 40 60 60 100 65.0 2 0 33.3 13.3 13.3 13.3 18.3 12 13.6 26.7 26.7 13.3 40 26.7 13 22.7 86.7 80 66.7 53.3 71.7 19 36.4 33.3 60 53.3 6.7 38.3 20 13.6 26.7 40 40 33.3 35.0 24 0 33.3 26.7 26.7 26.7 28.4 25 22.7 25.7 26.7 33.3 40 31.4 mean 14.2 38.2 41.738.3 39.2 39.3 sem 4.5 7.1 8.07.2 10.2 6.7
Statistical Analyses
Test: Between-groups comparison of context freezing
Number of Groups: 4 Number of Measurements: 1
304 Number of subjects in... Group 1 (Control): 7 Group 2 (Block - 0 mg/kg): 7 Group 3 (Block - 1 mg/kg): 8 Group 4 (Block - 10 mg/kg): 8
Between contrast coefficients Contrast Group... 1 2 3 4 B1 3 -1 -1 -1 B2 0 2 -1 -1 B3 0 1 0 -1 B4 0 1 -1 0 B5 0 0 1 -1 B6 0 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 10.371 Measurement 1 Mean 10.371 SD 10.709
Group 2 Overall Mean: 15.586 Measurement 1 Mean 15.586 SD 13.360
Group 3 Overall Mean: 11.338 Measurement 1 Mean 11.338 SD 11.917
305 Group 4 Overall Mean: 14.188 Measurement 1 Mean 14.188 SD 12.736
Means and SDs averaged across groups Measurement 1 Mean 12.871 SD 12.230 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 59.532 1 59.532 0.398 B2 38.813 1 38.813 0.260 B3 7.299 1 7.299 0.049 B4 67.377 1 67.377 0.450 B5 32.490 1 32.490 0.217 B6 2.745 1 2.745 0.018 Error 3888.650 26 149.563 ------
Test: Comparison on context freezing and average CS freezing
Number of Groups: 4 Number of Measurements: 5
306 Number of subjects in... Group 1 (Control): 7 Group 2 (Block - 0 mg/kg): 7 Group 3 (Block - 1 mg/kg): 8 Group 4 (Block - 10 mg/kg: 8
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 W1 4 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 58.074 Measurement 1 2 3 4 5 Mean 10.371 65.714 76.171 67.643 70.471 SD 10.709 28.404 26.080 25.650 20.685
Group 2 Overall Mean: 30.926 Measurement 1 2 3 4 5 Mean 15.586 39.029 43.814 32.371 23.829 SD 13.360 22.593 37.697 34.728 23.370
Group 3 Overall Mean: 38.935 Measurement 1 2 3 4 5 Mean 11.338 40.850 54.175 43.325 44.988 SD 11.917 21.803 18.996 28.728 16.245
Group 4 Overall Mean: 34.313 Measurement 1 2 3 4 5 Mean 14.188 38.213 41.675 38.325 39.163 SD 12.736 20.172 22.744 20.406 28.822
307 Means and SDs averaged across groups Measurement 1 2 3 4 5 Mean 12.871 45.951 53.959 45.416 44.613 SD 12.230 23.270 26.858 27.649 22.792 ------
Analysis of Variance Summary Table Source SS df MS F ------Between 38904.656 26 1496.333 ------Within ------W1 28627.583 1 28627.583 95.684 Error 7778.917 26 299.189 ------
Test: CS freezing
Number of Groups: 4 Number of Measurements: 4
Number of subjects in... Group 1 (Control): 7 Group 2 (Block - 0 mg/kg): 7 Group 3 (Block - 1 mg/kg): 8 Group 4 (Block - 10 mg/kg): 8
308 Between contrast coefficients Contrast Group... 1 2 3 4 B1 3 -1 -1 -1 B2 0 1 1 -2 B3 0 1 -1 0
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 -3 -1 1 3
Means and Standard Deviations Group 1 Overall Mean: 70.000 Measurement 1 2 3 4 Mean 65.714 76.171 67.643 70.471 SD 28.404 26.080 25.650 20.685
Group 2 Overall Mean: 34.761 Measurement 1 2 3 4 Mean 39.029 43.814 32.371 23.829 SD 22.593 37.697 34.728 23.370
Group 3 Overall Mean: 45.834 Measurement 1 2 3 4 Mean 40.850 54.175 43.325 44.988 SD 21.803 18.996 28.728 16.245
Group 4 Overall Mean: 39.344 Measurement 1 2 3 4 Mean 38.213 41.675 38.325 39.163 SD 20.172 22.744 20.406 28.822
309 Means and SDs averaged across groups Measurement 1 2 3 4 Mean 45.951 53.959 45.416 44.613 SD 23.270 26.858 27.649 22.792 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 19328.371 1 19328.371 11.743 B2 18.956 1 18.956 0.012 B3 1831.214 1 1831.214 1.113 Error 42794.923 26 1645.959 ------Within ------W1 235.555 1 235.555 0.622 B1W1 159.646 1 159.646 0.421 B2W1 193.272 1 193.272 0.510 B3W1 641.123 1 641.123 1.693 Error 9848.391 26 378.784 ------
310 Appendix 8: Experiment 7. Raw Data Conditioning (Day 1) Group 0 mg/kg Group 1 mg/kg rat context CS1 CS2 rat context CS1 CS2 1 0 2066.7 17 4.5 0 66.7 2 0 0 66.7 18 4.5 60 60 5 0 0 100 19 4.5 80 0 8 4.5 20 73.3 20 0 6.7 100 10 0 26.7100 21 0 13.353.3 12 0 0 86.7 22 0 0 86.7 14 0 6.773.3 23 0 0 86.7 15 0 6.773.3 mean 1.9 22.9 64.8 mean 0.6 10.0 80.0 sem 0.9 12.5 12.5 sem 0.6 3.8 4.9
group 10 mg/kg rat context CS1 CS2 3 4.5 0 93.3 4 0 0 66.7 6 0 6.7 33.3 7 0 13.3 93.3 9 0 0 93.3 11 0 0 60 13 0 13.3 100 16 0 0 100 mean 0.6 4.280.0 sem 0.6 2.28.5
Test/Extinction (Day 2) group 0 mg/kg rat context CS1 CS2 CS3 CS4 CS5 CS6 1 40.9 60 66.7 73.3 80 86.7 80 2 0 60 80 80 80 73.3 53.3 5 9.1 66.7 60 20 33.3 0 6.7 8 0 40 20 20 0 53.3 66.7 10 9.1 93.386.7 93.3 100 93.3 100 12 0 73.393.3 100 100 60 0 14 0 86.7 73.3 80 93.3 93.3 66.7 15 4.5 66.7 100 86.7 93.3 73.3 60 mean 8.0 68.3 72.5 69.2 72.5 66.7 54.2 sem 4.9 5.9 8.9 11.1 12.9 10.8 12.2
311 group 1 mg/kg rat context CS1 CS2 CS3 CS4 CS5 CS6 17 4.5 80 86.7 80 60 46.7 26.7 18 22.7 46.7 86.7 66.7 86.7 66.7 33.3 19 9.1 66.7 60 100 93.3 46.7 46.7 20 4.5 80 73.313.3 33.3 53.3 6.7 21 4.5 66.7 46.7 26.7 86.7 40 33.3 22 54.5 100 100 86.7 100 100 33.3 23 81.8 93.3 100 93.3 80 73.3 80 mean 25.9 76.2 79.1 66.7 77.1 61.0 37.1 sem 11.6 6.8 7.6 12.8 8.7 7.9 8.5
group 10 mg/kg rat context CS1 CS2 CS3 CS4 CS5 CS6 3 22.7 80 33.3 20 60 80 86.7 4 0 80 80 80 80 53.3 53.3 6 0 60 80 66.7 40 26.7 40 7 54.5 100 100 100 100 73.3 73.3 9 36.4 93.3 86.7 53.3 86.7 40 66.7 11 0 80 46.780 46.7 20 66.7 13 13.6 93.3 100 100 86.7 80 53.3 16 0 40 53.326.7 13.3 66.7 26.7 mean 15.9 78.3 72.5 65.8 64.2 55.0 58.3 sem 7.3 7.0 8.9 10.8 10.4 8.4 6.8
Test/Extinction (Day 3) group 0 mg/kg rat context CS1 CS2 CS3 CS4 CS5 CS6 1 13.6 66.7 33.3 73.3 66.7 53.3 53.3 2 0 53.3 6.746.7 33.3 40 6.7 5 0 0 0 0 0 0 13.3 8 0 0 26.7 0 60 66.7 93.3 10 4.5 86.7 80 13.3 66.7 46.7 80 12 0 0 33.3 0 13.3 6.7 13.3 14 0 86.7 8033.3 50 26.7 13.3 15 0 66.7 80 80 33.3 53.3 60 mean 2.3 45.0 42.5 30.8 40.4 36.7 41.7 sem 1.7 13.7 11.7 11.7 8.8 8.4 12.1
group 1 mg/kg rat context CS1 CS2 CS3 CS4 CS5 CS6 17 0 80 60 86.7 80 60 66.7 18 9.1 60 60 60 86.7 26.7 6.7 19 0 26.7 66.7 73.3 26.7 20 0 20 0 0 13.333.3 26.7 26.7 20 21 0 53.3 13.3 26.7 60 33.3 0 22 22.7 73.3 66.7 6.7 46.7 40 20 23 18.2 100 100 73.3 46.7 33.3 40 mean 7.1 56.2 54.3 51.4 53.4 34.3 21.9 sem 3.7 12.8 11.8 11.1 9.0 4.9 9.2
312 group 10 mg/kg rat context CS1 CS2 CS3 CS4 CS5 CS6 3 0 53.346.7 53.3 53.3 86.7 60 4 0 66.7 46.7 66.7 60 86.7 53.3 6 0 13.36.7 20 0 0 0 7 9.1 73.373.3 66.7 0 0 20 9 0 8013.3 40 13.3 20 46.7 11 0 0 0 33.3 13.3 40 53.3 13 31.8 100 93.3 46.7 66.7 60 53.3 16 0 0 0 6.7 0 40 0 mean 5.1 48.3 35.0 41.7 25.8 41.7 35.8 sem 4.0 13.7 12.6 7.5 10.3 12.2 8.9
Statistical Analyses
Conditioning: context freezing
Number of Groups: 3 Number of Measurements: 1
Number of subjects in... Group 1 (0 mg/kg): 8 Group 2 (1 mg/kg): 7 Group 3 (10 mg/kg): 8
Between contrast coefficients Contrast Group... 1 2 3 B1 1 -1 0 B2 1 0 -1 B3 0 1 -1
Means and Standard Deviations Group 1 Overall Mean: 0.563 Measurement 1 Mean 0.563 SD 1.591
313 Group 2 Overall Mean: 1.929 Measurement 1 Mean 1.929 SD 2.405
Group 3 Overall Mean: 0.563 Measurement 1 Mean 0.563 SD 1.591
Means and SDs averaged across groups Measurement 1 Mean 1.018 SD 1.873 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 6.967 1 6.967 1.986 B2 0.000 1 0.000 0.000 B3 6.967 1 6.967 1.986 Error 70.152 20 3.508 ------
Conditioning: CS freezing
Number of Groups: 3 Number of Measurements: 2
314 Number of subjects in... Group 1 (0 mg/kg): 8 Group 2 (1 mg/kg): 7 Group 3 (10 mg/kg): 8
Between contrast coefficients Contrast Group... 1 2 3 B1 1 -1 0 B2 1 0 -1 B3 0 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 45.006 Measurement 1 2 Mean 10.013 80.000 SD 10.695 13.801
Group 2 Overall Mean: 43.814 Measurement 1 2 Mean 22.857 64.771 SD 33.077 33.056
Group 3 Overall Mean: 42.075 Measurement 1 2 Mean 4.163 79.988 SD 6.095 24.167
315 Means and SDs averaged across groups Measurement 1 2 Mean 12.344 74.920 SD 19.526 24.472 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 10.608 1 10.608 0.046 B2 68.738 1 68.738 0.299 B3 22.588 1 22.588 0.098 Error 4591.772 20 229.589 ------Within ------W1 44852.622 1 44852.622 59.757 B1W1 1471.130 1 1471.130 1.960 B2W1 68.153 1 68.153 0.091 B3W1 2146.548 1 2146.548 2.860 Error 15011.576 20 750.579 ------
Test/Extinction: Context freezing
Number of Groups: 3 Number of Measurements: 2
Number of subjects in... Group 1 (0 mg/kg): 8 Group 2 (1 mg/kg): 7 Group 3 (10 mg/kg): 8
316 Between contrast coefficients Contrast Group... 1 2 3 B1 1 -1 0 B2 1 0 -1 B3 0 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 5.106 Measurement 1 2 Mean 7.950 2.263 SD 13.903 4.844
Group 2 Overall Mean: 16.543 Measurement 1 2 Mean 25.943 7.143 SD 30.575 9.766
Group 3 Overall Mean: 10.506 Measurement 1 2 Mean 15.900 5.113 SD 20.609 11.244
Means and SDs averaged across groups Measurement 1 2 Mean 16.598 4.839 SD 22.288 9.004 ------
317 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 976.610 1 976.610 2.398 B2 233.280 1 233.280 0.573 B3 272.090 1 272.090 0.668 Error 8145.033 20 407.252 ------Within ------W1 1583.687 1 1583.687 9.284 B1W1 320.950 1 320.950 1.882 B2W1 52.020 1 52.020 0.305 B3W1 119.840 1 119.840 0.703 Error 3411.529 20 170.576 ------
Test/Extinction: CS freezing
Number of Groups: 3 Number of Measurements: 12 Number of subjects in... Group 1 (0 mg/kg): 8 Group 2 (1 mg/kg): 7 Group 3 (10 mg/kg): 8
Between contrast coefficients Contrast Group... 1 2 3 B1 1 -1 0 B2 1 0 -1 B3 0 1 -1
318 Within contrast coefficients Contrast Measurement... 1 2 3 4 5 6 7 8 9 10 11 12 W1 -11 -9 -7 -5 -3 -1 1 3 5 7 9 11
Means and Standard Deviations Group 1 Overall Mean: 53.366 Measurement 1 2 3 4 5 6 7 8 Mean 68.338 72.500 69.163 72.488 66.650 54.175 45.013 42.500 SD 16.615 25.053 31.458 36.417 30.649 34.443 38.847 33.221 9 10 11 12 30.825 40.413 36.675 41.650 33.033 24.863 23.631 34.317
Group 2 Overall Mean: 55.719 Measurement 1 2 3 4 5 6 7 8 Mean 76.200 79.057 66.671 77.143 60.957 37.143 56.186 54.286 SD 17.974 20.155 33.776 23.059 20.873 22.391 33.736 31.153 9 10 11 12 51.429 53.357 34.286 21.914 29.482 23.719 13.006 24.263
Group 3 Overall Mean: 51.875 Measurement 1 2 3 4 5 6 7 8 Mean 78.325 72.500 65.838 64.175 55.000 58.338 48.325 35.000 SD 19.753 25.064 0.533 29.384 23.831 19.109 38.835 35.580 9 10 11 12 41.675 25.825 41.675 35.825 21.310 29.052 34.516 25.174
319 Means and SDs averaged across groups Measurement 1 2 3 4 5 6 7 8 Mean 74.288 74.686 67.224 71.268 60.869 49.885 49.841 43.929 SD 18.169 23.694 31.858 30.428 25.657 26.333 37.383 33.474 9 10 11 12 41.310 39.865 37.545 33.130 28.313 26.087 25.752 28.471 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 248.129 1 248.129 0.058 B2 106.654 1 106.654 0.025 B3 661.996 1 661.996 0.156 Error 85127.410 20 4256.371 ------Within ------W1 55688.582 1 55688.582 37.084 B1W1 87.320 1 87.320 0.058 B2W1 70.438 1 70.438 0.047 B3W1 1.529 1 1.529 0.001 Error 30033.604 20 1501.680 ------
320 Appendix 9: Experiment 8a Raw Data Conditioning (Number after group name refers to conditioning context) Same - 1 Same - 2 rat context CS1 CS2 rat context CS1 CS2 1 4.5 0 80 5 4.5 33.3 66.7 2 0 73.3 66.7 6 0 26.7 60 15 4.5 33.3 20 11 0 66.7 100 16 0 13.3 53.3 12 0 0 93.3 mean 2.3 30.055.0 mean 1.1 31.780.0
Different - 1 Different - 2 rat context CS1 CS2 rat context CS1 CS2 3 22.7 20 86.7 7 0 6.7 53.3 4 0 20 46.7 8 0 46.7 73.3 13 0 0 13.3 9 9.1 0 93.3 14 0 26.7 93.3 10 0 0 80 mean 5.7 16.760.0 mean 2.3 13.475.0
Test Same - 1 rat context CS1 CS2 CS3 CS4 CS mean 1 59.1 100 40 93.3 93.3 81.7 2 13.6 46.7 80 80 86.7 73.4 15 54.5 33.3 60 53.3 86.7 58.3 16 63.6 100 100 100 20 80.0 mean 47.7 70.0 70.0 81.7 71.7 73.3
Same - 2 rat context CS1 CS2 CS3 CS4 CS mean 5 0 86.7 86.7 66.7 66.7 76.7 6 40.9 73.3 73.3 40 40 56.7 11 27.3 53.3 100 93.3 33.3 70.0 12 13.6 86.7 66.7 86.7 86.7 81.7 mean 20.5 75.0 81.7 71.7 56.7 71.3
Groups Same mean 34.1 72.5 75.8 76.7 64.2 72.3 sem 8.5 9.0 7.2 7.6 10.2 3.5
Different - 1 rat context CS1 CS2 CS3 CS4 CS mean 3 13.6 60 46.7 66.7 53.3 56.7 4 13.6 80 66.7 73.3 53.3 68.3 13 0 46.7 80 93.3 86.7 76.7 14 4.5 100 93.3 100 26.7 80.0 mean 7.9 71.7 71.7 83.3 55.0 70.4
321 Different - 2 rat context CS1 CS2 CS3 CS4 CS mean 7 0 60 93.380 20 63.3 8 9.1 93.3 80 86.7 53.3 78.3 9 31.8 86.7 80 66.7 13.3 61.7 10 9.1 93.3 86.793.3 80 88.3 mean 12.5 83.3 85.0 81.7 41.7 72.9
Groups Different mean 10.2 77.5 78.3 82.5 48.3 71.7 sem 3.6 6.9 5.4 4.5 9.5 3.8
Statistical Analyses
Test: Between-groups and between-contexts comparisons of context freezing
Number of Groups: 4 Number of Measurements: 1
Number of subjects in... Group 1 (Same - 1): 4 Group 2 (Same - 2): 4 Group 3 (Different - 1): 4 Group 4 (Different - 2): 4
Between contrast coefficients Contrast Group... 1 2 3 4 B1 1 1 -1 -1 B2 1 -1 1 -1 B3 1 -1 -1 1
Means and Standard Deviations Group 1 Overall Mean: 47.700 Measurement 1 Mean 47.700 SD 23.035
322 Group 2 Overall Mean: 20.450 Measurement 1 Mean 20.450 SD 17.609
Group 3 Overall Mean: 7.925 Measurement 1 Mean 7.925 SD 6.806
Group 4 Overall Mean: 12.500 Measurement 1 Mean 12.500 SD 13.563
Means and SDs averaged across groups Measurement 1 Mean 22.144 SD 16.363 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2277.676 1 2277.676 8.507 B2 514.156 1 514.156 1.920 B3 1012.831 1 1012.831 3.783 Error 3212.878 12 267.740 ------
323 Test: Comparison of context freezing and mean CS freezing
Number of Groups: 4 Number of Measurements: 5
Number of subjects in... Group 1 (Same - 1): 4 Group 2 (Same - 2): 4 Group 3 (Different - 1): 4 Group 4 (Different - 2): 4
Between contrast coefficients Contrast Group... 1 2 3 4 B1 1 1 -1 -1 B2 1 -1 1 -1 B3 1 -1 -1 1
Within contrast coefficients Contrast Measurement... 1 2 3 4 5 W1 4 -1 -1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 68.205 Measurement 1 2 3 4 5 Mean 47.700 70.000 70.000 81.650 71.675 SD 23.035 35.070 25.820 20.647 34.590
Group 2 Overall Mean: 61.095 Measurement 1 2 3 4 5 Mean 20.450 75.000 81.675 71.675 56.675 SD 17.609 15.786 14.781 23.954 24.674
324 Group 3 Overall Mean: 57.920 Measurement 1 2 3 4 5 Mean 7.925 71.675 71.675 83.325 55.000 SD 6.806 23.321 19.878 15.858 24.573
Group 4 Overall Mean: 60.830 Measurement 1 2 3 4 5 Mean 12.500 83.325 85.000 81.675 41.650 SD 13.563 15.858 6.371 11.364 30.978
Means and SDs averaged across groups Measurement 1 2 3 4 5 Mean 22.144 75.000 77.088 79.581 56.250 SD 16.363 23.846 18.172 18.579 29.021 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 556.512 1 556.512 1.562 B2 88.200 1 88.200 0.247 B3 502.002 1 502.002 1.409 Error 4276.729 12 356.394 ------Within ------W1 31790.345 1 31790.345 77.171 B1W1 1727.476 1 1727.476 4.193 B2W1 426.657 1 426.657 1.036 B3W1 594.323 1 594.323 1.443 Error 4943.387 12 411.949 ------
325 Test: CS freezing
Number of Groups: 4 Number of Measurements: 4
Number of subjects in... Group 1 (Same - 1): 4 Group 2 (Same - 2): 4 Group 3 (Different - 1): 4 Group 4 (Different - 2): 4
Between contrast coefficients Contrast Group... 1 2 3 4 B1 1 1 -1 -1 B2 1 -1 1 -1 B3 1 -1 -1 1
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 -3 -1 1 3
Means and Standard Deviations Group 1 Overall Mean: 73.331 Measurement 1 2 3 4 Mean 70.000 70.000 81.650 71.675 SD 35.070 25.820 20.647 34.590
Group 2 Overall Mean: 71.256 Measurement 1 2 3 4 Mean 75.000 81.675 71.675 56.675 SD 15.786 14.781 23.954 24.674
326 Group 3 Overall Mean: 70.419 Measurement 1 2 3 4 Mean 71.675 71.675 83.325 55.000 SD 23.321 19.878 15.858 24.573
Group 4 Overall Mean: 72.913 Measurement 1 2 3 4 Mean 83.325 85.000 81.675 41.650 SD 15.858 6.371 11.364 30.978
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 75.000 77.088 79.581 56.250 SD 23.846 18.172 18.579 29.021 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 6.313 1 6.313 0.013 B2 0.701 1 0.701 0.001 B3 83.494 1 83.494 0.167 Error 6007.238 12 500.603 ------Within ------W1 2311.788 1 2311.788 2.909 B1W1 701.224 1 701.224 0.882 B2W1 1472.757 1 1472.757 1.853 B3W1 3.465 1 3.465 0.004 Error 9536.663 12 794.722 ------
327 Appendix 10: Experiment 8b Raw Data Test (Number after group name refers to conditioning context) Same - 1 Different – 1 rat period 1 period 2 Rat period 1 period 2 1 26.7 66.7 3 0 6.7 2 86.7 73.3 4 0 6.7 15 20 66.7 13 66.7 0 16 26.7 60 14 80 53.3 mean 40.0 66.7 Mean 36.7 16.7
Same - 2 Different – 2 rat period 1 period 2 Rat period 1 period 2 5 60 60 7 86.7 40 6 80 86.7 8 13.3 6.7 11 40 53.3 9 46.7 33.3 12 0 0 10 66.7 46.7 mean 45.0 50.0 Mean 53.4 31.7
Groups Groups Same Different mean 42.5 58.3 Mean 45.0 24.2 sem 10.8 9.1 Sem 12.7 7.5
Statistical Analyses
Test: Context Freezing
Number of Groups: 4 Number of Measurements: 2
Number of subjects in... Group 1 (Same - 1): 4 Group 2 (Same - 2): 4 Group 3 (Different - 1): 4 Group 4 (Different - 2): 4
328 Between contrast coefficients Contrast Group... 1 2 3 4 B1 1 1 -1 -1 B2 1 -1 1 -1 B3 1 -1 -1 1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 53.350 Measurement 1 2 Mean 40.025 66.675 SD 31.277 5.430
Group 2 Overall Mean: 47.500 Measurement 1 2 Mean 45.000 50.000 SD 34.157 36.322
Group 3 Overall Mean: 26.675 Measurement 1 2 Mean 36.675 16.675 SD 42.695 24.620
Group 4 Overall Mean: 42.513 Measurement 1 2 Mean 53.350 31.675 SD 31.298 17.526
329 Means and SDs averaged across groups Measurement 1 2 Mean 43.763 41.256 SD 35.169 23.780 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2005.028 1 2005.028 1.325 B2 199.500 1 199.500 0.132 B3 940.695 1 940.695 0.621 Error 18164.839 12 1513.737 ------Within ------W1 50.250 1 50.250 0.174 B1W1 2688.278 1 2688.278 9.315 B2W1 272.028 1 272.028 0.943 B3W1 199.500 1 199.500 0.691 Error 3463.289 12 288.607 ------
Test: Between-groups comparison of context freezing during first 2 min period
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Same): 8 Group 2 (Different): 8
330 Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 42.513 Measurement 1 Mean 42.513 SD 30.435
Group 2 Overall Mean: 45.013 Measurement 1 Mean 45.013 SD 35.784
Means and SDs averaged across groups Measurement 1 Mean 43.763 SD 33.217 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 25.000 1 25.000 0.023 Error 15447.618 14 1103.401 ------
331 Test: Between-groups comparison of context freezing during second 2 min period
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Same): 8 Group 2 (Different): 8
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 58.338 Measurement 1 Mean 58.338 SD 25.641
Group 2 Overall Mean: 24.175 Measurement 1 Mean 24.175 SD 21.347
Means and SDs averaged across groups Measurement 1 Mean 41.256 SD 23.592 ------
332 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 4668.306 1 4668.306 8.387 Error 7792.234 14 556.588 ------
333 Appendix 11: Experiment 9 Raw Data Stage I conditioning (Number after group name refers to conditioning context.) Tone Control (Same - 1) Day 1 Day 3 rat context CS context CS 4 0 60 0 40 8 0 26.7 0 86.7 15 0 0 31.8 60 21 0 33.3 0 53.3 23 0 13.3 9.1 60 29 0 33.3 31.8 33.3 38 0 0 45.5 80 Mean 0.0 23.8 16.959.0
Control (Same - 2) Day 1 Day 3 rat context CS context CS 1 22.7 0 45.5 80 10 9.1 26.7 18.2 33.3 16 0 0 59.1 53.3 17 18.2 26.7 40.9 93.3 36 0 26.7 0 46.7 46 0 0 9.1 86.7 Mean 8.3 13.4 28.865.6
Control (Different - 1) Day 1 Day 3 rat context CS context CS 33 4.5 20 0 33.3 37 0 6.7 72.7 73.3 Mean 2.3 13.4 36.453.3
Control (Different - 2) Day 1 Day 3 rat context CS context CS 27 36.4 46.7 68.2 40 41 0 0 40.9 73.3 44 0 0 13.6 46.7 Mean 12.1 15.6 40.953.3
Groups Control Mean 5.1 17.8 27.059.6 sem 2.4 4.3 5.94.8
334 AAA - 1 Day 1 Day 3 rat context CS context CS 5 0 66.7 4.5 40 9 0 0 13.6 66.7 13 0 6.7 0 60 24 0 20 9.1 33.3 Mean 0.0 23.4 6.850.0
AAA - 2 Day 1 Day 3 rat context CS context CS 2 9.1 26.7 4.5 60 12 0 0 18.2 73.3 20 0 13.3 4.5 33.3 Mean 3.0 13.3 9.155.5
ABB - 1 Day 1 Day 3 rat context CS context CS 28 0 13.3 4.5 93.3 31 0 13.3 22.7 66.7 40 0 20 54.5 33.3 47 0 33.3 13.3 33.3 Mean 0.0 20.0 23.856.7
ABB - 2 Day 1 Day 3 rat context CS context CS 25 4.5 6.7 54.5 66.7 35 4.5 0 0 53.3 43 4.5 0 18.2 93.3 Mean 4.5 2.2 24.271.1
Groups Same Mean 1.6 15.7 15.957.6 sem 0.8 4.8 4.85.6
ABA - 1 Day 1 Day 3 rat context CS context CS 3 0 53.3 27.3 40 6 0 53.3 9.1 80 7 4.5 40 22.7 66.7 14 31.8 86.7 0 73.3 22 0 0 0 93.3 30 0 80 18.2 33.3 32 0 13.3 13.6 40 34 0 13.3 9.1 60 39 0 26.7 4.5 46.7 Mean 4.0 40.7 11.659.3
335 ABA - 2 Day 1 Day 3 rat context CS context CS 11 9.1 46.7 68.2 66.7 18 0 0 81.8 33.3 19 0 53.3 0 40 26 13.6 6.7 0 60 42 9.1 13.3 22.7 93.3 45 0 6.7 36.4 46.7 Mean 5.3 21.1 34.956.7
Groups Different Mean 4.5 32.9 20.958.2 sem 2.3 7.3 6.45.3
Flash Control (Same - 1) Day 2 Day 4 rat context CS context CS 4 4.5 60 4.5 73.3 8 31.8 26.7 10 80 15 4.5 13.3 18.2 66.7 21 0 20 40.9 80 23 4.5 6.7 4.5 80 29 18.2 28.6 4.5 53.3 38 59.1 73.3 0 86.7 Mean 17.5 32.7 11.874.3
Control (Same - 2) Day 2 Day 4 rat context CS context CS 1 4.5 53.3 26.7 66.7 10 40.9 26.7 4.5 53.3 16 18.2 60 31.8 33.3 17 31.8 86.7 40.9 86.7 36 9.1 26.7 0 66.7 46 0 86.7 4.5 86.7 Mean 17.4 56.7 18.165.6
Control (Different - 1) Day 2 Day 4 rat context CS context CS 33 0 0 0 53.3 37 4.5 40 22.7 66.7 Mean 2.3 20.0 11.460.0
336 Control (Different - 2) Day 2 Day 4 rat context CS context CS 27 72.7 60 9.1 93.3 41 9.1 86.7 20 53.3 44 13.6 86.7 4.5 13.3 Mean 31.8 77.8 11.253.3
Groups Control Mean 18.2 46.8 13.766.3 sem 5.0 7.0 3.24.8
AAA - 1 Day 2 Day 4 rat context CS context CS 5 0 20 4.5 86.7 9 4.5 26.7 6.7 53.3 13 0 53.3 0 93.3 24 0 6.7 4.5 53.3 Mean 1.1 26.7 3.971.7
AAA - 2 Day 2 Day 4 rat context CS context CS 2 4.5 26.7 0 93.3 12 9.1 80 36.4 13.3 20 0 6.7 9.1 20 Mean 4.5 37.8 15.242.2
ABB - 1 Day 2 Day 4 rat context CS context CS 28 9.1 64.3 13.6 66.7 31 36.4 13.3 4.5 46.7 40 13.6 6.7 9.1 66.7 47 23.3 13.3 3.3 6.7 Mean 20.6 24.4 7.646.7
ABB - 2 Day 2 Day 4 rat context CS context CS 25 18.2 33.3 45.5 60 35 0 60 24.4 80 43 22.7 73.3 3.3 73.3 Mean 13.6 55.5 24.471.1
Groups Same Mean 10.1 34.6 11.858.1 sem 3.0 7.0 3.77.6
337 ABA - 1 Day 2 Day 4 rat context CS context CS 3 4.5 33.3 4.5 73.3 6 0 26.7 13.6 100 7 45.5 33.3 26.7 66.7 14 22.7 13.3 4.5 80 22 0 13.3 0 66.7 30 0 0 0 26.7 32 0 80 0 86.7 34 13.6 0 0 13.3 39 4.5 6.7 0 20 Mean 10.1 23.0 5.559.3
ABA - 2 Day 2 Day 4 rat context CS context CS 11 0 6.7 0 66.7 18 0 86.7 72.7 73.3 19 9.1 60 0 86.7 26 18.2 93.3 9.1 60 42 0 6.7 46.7 66.7 45 72.7 86.7 63.6 86.7 Mean 16.7 56.7 32.073.4
Groups Different Mean 12.7 36.4 16.164.9 sem 5.4 9.1 6.46.6
338 Stage II conditioning AAA - 1 ABA - 1 rat context CS1 CS2 rat context CS1 CS2 5 36.4 66.7 26.7 3 3.3 26.7 0 9 18.2 66.7 40 6 23.3 80 20 13 0 46.7 66.7 7 0 86.7 80 24 0 40 26.7 14 0 46.7 26.7 Mean 13.7 55.0 40.0 22 0 53.326.7 30 0 33.3 20 AAA - 2 32 0 73.3 46.7 rat context CS1 CS2 34 0 46.7 20 2 0 80 26.7 39 3.3 66.7 26.7 12 40.9 66.7 13.3 Mean 3.3 57.0 29.6 20 3.3 0 0 Mean 14.7 48.9 13.3 ABA - 2 rat context CS1 CS2 ABB - 1 11 0 46.7 26.7 rat context CS1 CS2 18 0 6.7 13.3 28 0 60 66.7 19 0 53.3 46.7 31 6.7 20 6.7 26 0 33.3 20 40 0 20 6.7 42 0 66.7 60 47 0 40 13.3 45 0 100 60 Mean 1.7 35.023.4 Mean 0.0 51.1 37.8
Groups ABB - 2 Different rat context CS1 CS2 Mean 2.0 54.7 32.9 25 0 80 80 sem 1.6 6.4 5.5 35 0 80 80 43 0 46.7 40 Mean 0.0 68.966.7
Groups Same Mean 7.5 51.035.3 sem 3.8 6.77.4
339 Test Control (Same - 1) rat context CS1 CS2 CS3 CS4 CS mean 4 0 40 0 6.7 13.3 15.0 8 13.3 86.7 100 100 73.3 90.0 15 0 66.7 66.7 60 60 63.4 21 0 86.7 86.7 93.3 80 86.7 23 0 60 73.3 73.3 73.3 70.0 29 0 40 60 46.7 53.3 50.0 38 6.7 80 86.7 26.7 40 58.4 Mean 2.9 65.767.6 58.1 56.2 61.9
Control (Same - 2) rat context CS1 CS2 CS3 CS4 CS mean 1 10 60 73.3 33.3 66.7 58.3 10 9.1 66.7 46.7 13.3 6.7 33.4 16 0 66.7 60 33.3 73.3 58.3 17 18.2 60 93.3 86.7 100 85.0 36 0 86.7 66.7 33.3 46.7 58.4 46 10 93.3 100 93.3 80 91.7 Mean 7.9 72.273.3 48.9 62.2 64.2
Control (Different - 1) rat context CS1 CS2 CS3 CS4 CS mean 33 0 46.7 13.3 6.7 0 16.7 37 22.7 53.3 26.7 66.7 46.7 48.4 Mean 11.4 50.020.0 36.7 23.4 32.5
Control (Different - 2) rat context CS1 CS2 CS3 CS4 CS mean 27 4.5 60 66.7 80 46.7 63.4 41 0.0 40.0 46.7 60.0 26.7 43.4 44 0 53.3 60 53.3 33.3 50.0 Mean 1.5 51.157.8 64.4 35.6 52.2
Groups Control context CS1 CS2 CS3 CS4 CS mean Mean 5.3 63.762.6 53.7 51.1 57.8 sem 1.7 4.06.6 7.2 6.5 5.3
340 AAA - 1 rat context CS1 CS2 CS3 CS4 CS mean 5 4.5 13.3 53.3 26.7 73.3 41.7 9 23.3 66.7 40 40 53.3 50.0 13 45.5 60 13.3 20 26.7 30.0 24 40.9 13.3 40 13.3 6.7 18.3 Mean 28.6 38.3 36.7 25.0 40.0 35.0
AAA - 2 rat context CS1 CS2 CS3 CS4 CS mean 2 20 20 20 26.7 40 26.7 12 36.4 13.3 66.7 40 26.7 36.7 20 4.5 33.3 20 20 0 18.3 Mean 20.3 22.2 35.6 28.9 22.2 27.2
ABB - 1 rat context CS1 CS2 CS3 CS4 CS mean 28 4.5 53.3 66.7 46.7 13.3 45.0 31 2.2 53.3 40 20 0 28.3 40 13.6 26.7 0 13.3 0 10.0 47 3.3 53.3 46.7 13.3 26.7 35.0 Mean 5.9 46.7 38.4 23.3 10.0 29.6
ABB - 2 rat context CS1 CS2 CS3 CS4 CS mean 25 31.8 86.7 100 60 46.7 73.4 35 0 66.7 66.7 60 46.7 60.0 43 10 46.7 40 20 53.3 40.0 Mean 13.9 66.7 68.9 46.7 48.9 57.8
Groups Same context CS1 CS2 CS3 CS4 CS mean Mean 17.2 43.3 43.8 30.0 29.5 36.7 sem 4.2 6.3 7.0 4.4 6.3 4.6
ABA - 1 rat context CS1 CS2 CS3 CS4 CS mean 3 4.5 20 0 6.7 13.3 10.0 6 0 33.3 20 46.7 73.3 43.3 7 16.7 86.7 80 26.7 53.3 61.7 14 9.1 66.7 66.7 20 80 58.4 22 0 26.7 40 13.3 20 25.0 30 0 13.3 20 20 20 18.3 32 6.7 46.7 26.7 13.3 26.7 28.4 34 6.7 46.7 33.3 20 26.7 31.7 39 13.3 93.3 86.7 73.3 93.3 86.7 Mean 6.3 48.2 41.5 26.7 45.2 40.4 341 ABA - 2 rat context CS1 CS2 CS3 CS4 CS mean 11 0 80 86.7 60 60 71.7 18 22.7 40 0 6.7 20 16.7 19 13.6 60 46.7 33.3 33.3 43.3 26 0 53.3 13.3 0 20 21.7 42 4.5 73.3 33.3 26.7 33.3 41.7 45 20 66.7 40 73.3 93.3 68.3 Mean 10.1 62.236.7 33.3 43.3 43.9
Groups Different context CS1 CS2 CS3 CS4 CS mean Mean 7.9 53.839.6 29.3 44.4 41.8 sem 2.0 6.37.5 6.1 7.4 6.0
Statistical Analyses
Stage I: Test for possible differences in CS freezing between different physical training contexts
Number of Groups: 2 Number of Measurements: 2
Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 60.127 Measurement 1 2 Mean 56.915 63.338 SD 20.071 24.387
342 Group 2 Overall Mean: 62.060 Measurement 1 2 Mean 60.629 63.490 SD 20.535 25.011
Means and SDs averaged across groups Measurement 1 2 Mean 58.772 63.414 SD 20.279 24.666 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 86.778 1 86.778 0.135 Error 28862.168 45 641.382 ------
Stage I: Test for possible differences in context freezing between different physical training contexts
Number of Groups: 2 Number of Measurements: 2 Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
343 Means and Standard Deviations Group 1 Overall Mean: 11.892 Measurement 1 2 Mean 16.062 7.723 SD 18.641 9.903
Group 2 Overall Mean: 25.174 Measurement 1 2 Mean 28.786 21.562 SD 26.090 22.199
Means and SDs averaged across groups Measurement 1 2 Mean 22.424 14.642 SD 22.261 16.538 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 4098.445 1 4098.445 7.646 Error 24121.043 45 536.023 ------
Stage I: Comparison of mean context freezing and mean CS freezing on Days 3 and 4
Number of Groups: 2 Number of Measurements: 4
Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
344 Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 4 W1 1 1 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 36.010 Measurement 1 2 3 4 Mean 16.062 7.723 56.915 63.338 SD 18.641 9.903 20.071 24.387
Group 2 Overall Mean: 43.617 Measurement 1 2 3 4 Mean 28.786 21.562 60.629 63.490 SD 26.090 22.199 20.535 25.011
Means and SDs averaged across groups Measurement 1 2 3 4 Mean 22.424 14.642 58.772 63.414 SD 22.261 16.538 20.279 24.666 ------
345 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2688.979 1 2688.979 4.230 Error 28605.932 45 635.687 ------Within ------W1 84170.785 1 84170.785 155.378 B1W1 1496.244 1 1496.244 2.762 Error 24377.280 45 541.717 ------
Stage II: Test for possible differences in context freezing based on different Stage I training physical context
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Context 1): 17 Group 2 (Context 2): 12
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
346 Means and Standard Deviations Group 1 Overall Mean: 5.365 Measurement 1 Mean 5.365 SD 10.537
Group 2 Overall Mean: 3.683 Measurement 1 Mean 3.683 SD 11.759
Means and SDs averaged across groups Measurement 1 Mean 4.524 SD 11.051 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 19.887 1 19.887 0.163 Error 3297.395 27 122.126 ------
Stage II: Test for possible differences in context freezing based on different Stage II physical training context
Number of Groups: 2 Number of Measurements: 1
347 Number of subjects in... Group 1 (Context 1): 13 Group 2 (Context 2): 16
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 4.200 Measurement 1 Mean 4.200 SD 10.904
Group 2 Overall Mean: 5.050 Measurement 1 Mean 5.050 SD 11.211
Means and SDs averaged across groups Measurement 1 Mean 4.625 SD 11.076 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 5.182 1 5.182 0.042 Error 3312.100 27 122.670 ------
348 Stage II: Comparison of context freezing between rats given Stage II training in the same context as Stage I and rats given Stage II training in a different context
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (AAA): 7 Group 2 (ABB and ABA): 22
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 14.114 Measurement 1 Mean 14.114 SD 18.007
Group 2 Overall Mean: 1.664 Measurement 1 Mean 1.664 SD 5.111
Means and SDs averaged across groups Measurement 1 Mean 7.889 SD 9.611 ------
349 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 823.203 1 823.203 8.912 Error 2494.079 27 92.373 ------
Stage II: Comparison of context freezing to mean CS freezing
Number of Groups: 2 Number of Measurements: 3 Number of subjects in... Group 1 (AAA): 7 Group 2 (ABA and ABB): 22
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 W1 2 -1 -1
Means and Standard Deviations Group 1 Overall Mean: 31.700 Measurement 1 2 3 Mean 14.114 52.400 28.586 SD 18.007 26.793 20.994
350 Group 2 Overall Mean: 30.156 Measurement 1 2 3 Mean 1.664 53.036 35.768 SD 5.111 24.383 25.411
Means and SDs averaged across groups Measurement 1 2 3 Mean 7.889 52.718 32.177 SD 9.611 24.939 24.498 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 37.976 1 37.976 0.052 Error 19614.223 27 726.453 ------Within ------W1 16912.345 1 16912.345 42.332 B1W1 947.549 1 947.549 2.372 Error 10786.866 27 399.514 ------
Stage II: Test for possible summation
Number of Groups: 2 Number of Measurements: 3 Number of subjects in... Group 1 (AAA): 7 Group 2 (ABB and ABA): 22
351 Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 W1 1 1 -2
Means and Standard Deviations Group 1 Overall Mean: 54.600 Measurement 1 2 3 Mean 52.371 59.029 52.400 SD 16.534 33.652 26.793
Group 2 Overall Mean: 58.388 Measurement 1 2 3 Mean 59.691 62.436 53.036 SD 21.426 24.991 24.383
Means and SDs averaged across groups Measurement 1 2 3 Mean 56.031 60.732 52.718 SD 20.440 27.156 24.939 ------
352 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 228.579 1 228.579 0.244 Error 25344.277 27 938.677 ------Within ------W1 454.237 1 454.237 1.262 B1W1 79.114 1 79.114 0.220 Error 9721.536 27 360.057 ------
Stage II: Test for possible differences in CS freezing based on different Stage I physical training context
Number of Groups: 2 Number of Measurements: 2 Number of subjects in... Group 1 (Context 1): 17 Group 2 (Context 2): 12
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
353 Means and Standard Deviations Group 1 Overall Mean: 40.994 Measurement 1 2 Mean 51.382 30.606 SD 20.247 22.619
Group 2 Overall Mean: 46.950 Measurement 1 2 Mean 55.008 38.892 SD 30.366 26.645
Means and SDs averaged across groups Measurement 1 2 Mean 53.195 34.749 SD 24.872 24.340 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 499.062 1 499.062 0.489 Error 27570.269 27 1021.121 ------Within ------W1 4787.330 1 4787.330 25.209 B1W1 76.373 1 76.373 0.402 Error 5127.524 27 189.908 ------
354 Stage II: Test for possible differences in CS freezing in different Stage II physical training contexts
Number of Groups: 2 Number of Measurements: 2 Number of subjects in... Group 1 (Context 1): 13 Group 2 (Context 2): 16
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 50.781 Measurement 1 2 Mean 56.423 45.138 SD 23.975 22.469
Group 2 Overall Mean: 37.509 Measurement 1 2 Mean 50.006 25.013 SD 25.301 22.383
Means and SDs averaged across groups Measurement 1 2 Mean 53.215 35.075 SD 24.720 22.421
355 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2526.553 1 2526.553 2.671 Error 25542.778 27 946.029 ------Within ------W1 4719.878 1 4719.878 28.132 B1W1 673.993 1 673.993 4.017 Error 4529.903 27 167.774 ------
Stage II: Test for possible differences in CS freezing based on whether or not there was a shift between Stage I and Stage II training context
Number of Groups: 2 Number of Measurements: 2 Number of subjects in... Group 1 (AAA): 7 Group 2 (ABB and ABA): 22
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
356 Means and Standard Deviations Group 1 Overall Mean: 40.493 Measurement 1 2 Mean 52.400 28.586 SD 26.793 20.994
Group 2 Overall Mean: 44.402 Measurement 1 2 Mean 53.036 35.768 SD 24.383 25.411
Means and SDs averaged across groups Measurement 1 2 Mean 52.718 32.177 SD 24.939 24.498 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 162.322 1 162.322 0.157 Error 27907.009 27 1033.593 ------Within ------W1 4481.318 1 4481.318 23.771 B1W1 113.778 1 113.778 0.604 Error 5090.118 27 188.523 ------
357 Stage II: Within-subjects change in CS freezing
Number of Groups: 1 Number of Measurements: 2 Number of subjects in... Group 1: 29
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 43.459 Measurement 1 2 Mean 52.883 34.034 SD 24.491 24.259
Analysis of Variance Summary Table Source SS df MS F ------Between 28069.331 28 1002.476 ------Within ------W1 5151.234 1 5151.234 27.717 Error 5203.896 28 185.853 ------
Test: Test for possible difference in context freezing based on physical context of Stage I training
Number of Groups: 2 Number of Measurements: 1
358 Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 9.135 Measurement 1 Mean 9.135 SD 12.237
Group 2 Overall Mean: 10.252 Measurement 1 Mean 10.252 SD 10.963
Means and SDs averaged across groups Measurement 1 Mean 9.693 SD 11.688 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 14.514 1 14.514 0.106 Error 6147.271 45 136.606 ------
359 Test: Test for possible differences in context freezing based on different physical test contexts
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 9.135 Measurement 1 Mean 9.135 SD 12.866
Group 2 Overall Mean: 10.252 Measurement 1 Mean 10.252 SD 10.022
Means and SDs averaged across groups Measurement 1 Mean 9.693 SD 11.688 ------
360 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 14.514 1 14.514 0.106 Error 6147.271 45 136.606 ------
Test: Test for possible difference in context freezing Control group rats tested in same vs different context to that used for Stage I training
Number of Groups: 2 Number of Measurements: 1 Number of subjects in... Group 1 (Same): 13 Group 2 (Different): 5
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 5.177 Measurement 1 Mean 5.177 SD 6.373
Group 2 Overall Mean: 5.440 Measurement 1 Mean 5.440 SD 9.843
361 Means and SDs averaged across groups Measurement 1 Mean 5.308 SD 7.395 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 0.250 1 0.250 0.005 Error 874.995 16 54.687 ------
Test: Test for possible differences in CS freezing between rats in Control group tested in same vs different context to that used for Stage I training
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Same): 13 Group 2 (Different): 5
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
362 Means and Standard Deviations Group 1 Overall Mean: 62.950 Measurement 1 Mean 62.950 SD 22.513
Group 2 Overall Mean: 44.360 Measurement 1 Mean 44.360 SD 17.145
Means and SDs averaged across groups Measurement 1 Mean 53.655 SD 21.298 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 1247.957 1 1247.957 2.751 Error 7257.998 16 453.625 ------
Test: Context freezing
Number of Groups: 3 Number of Measurements: 1
363 Number of subjects in... Group 1 (Control): 18 Group 2 (Same): 14 Group 3 (Different): 15
Between contrast coefficients Contrast Group... 1 2 3 B1 2 -1 -1 B2 0 1 -1
Means and Standard Deviations Group 1 Overall Mean: 5.250 Measurement 1 Mean 5.250 SD 7.175
Group 2 Overall Mean: 17.179 Measurement 1 Mean 17.179 SD 15.817
Group 3 Overall Mean: 7.853 Measurement 1 Mean 7.853 SD 7.764
Means and SDs averaged across groups Measurement 1 Mean 10.094 SD 10.629 ------
364 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 586.084 1 586.084 5.187 B2 629.711 1 629.711 5.573 Error 4971.386 44 112.986 ------
Test: Context freezing with group Same split into groups AAA and ABB
Number of Groups: 4 Number of Measurements: 1
Number of subjects in... Group 1 (Control): 18 Group 2 (AAA): 7 Group 3 (ABB): 7 Group 4 (Different): 15
Between contrast coefficients Contrast Group... 1 2 3 4 B1 0 2 -1 -1 B2 0 0 1 -1
Means and Standard Deviations Group 1 Overall Mean: 5.250 Measurement 1 Mean 5.250 SD 7.175
365 Group 2 Overall Mean: 25.014 Measurement 1 Mean 25.014 SD 16.690
Group 3 Overall Mean: 9.343 Measurement 1 Mean 9.343 SD 10.963
Group 4 Overall Mean: 7.853 Measurement 1 Mean 7.853 SD 7.764
Means and SDs averaged across groups Measurement 1 Mean 11.865 SD 9.779 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 1380.321 1 1380.321 14.435 B2 10.589 1 10.589 0.111 Error 4111.808 43 95.623 ------
366 Test: Comparison of context freezing to mean CS freezing
Number of Groups: 3 Number of Measurements: 2 Number of subjects in... Group 1 (Control): 18 Group 2 (Same): 14 Group 3 (Different): 15
Within contrast coefficients Contrast Measurement... 1 2 W1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 31.528 Measurement 1 2 Mean 5.250 57.806 SD 7.175 22.371
Group 2 Overall Mean: 26.925 Measurement 1 2 Mean 17.179 36.671 SD 15.817 17.051
Group 3 Overall Mean: 24.823 Measurement 1 2 Mean 7.853 41.793 SD 7.764 23.068
Means and SDs averaged across groups Measurement 1 2 Mean 10.094 45.423 SD 10.629 21.180
367 Analysis of Variance Summary Table Source SS df MS F ------Between 13825.068 44 314.206 ------Within ------W1 29004.641 1 29004.641 117.261 Error 10883.465 44 247.351 ------
Test: Test for possible difference in CS freezing based on physical Stage I training context
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 43.079 Measurement 1 Mean 43.079 SD 23.751
368 Group 2 Overall Mean: 50.480 Measurement 1 Mean 50.480 SD 21.177
Means and SDs averaged across groups Measurement 1 Mean 46.779 SD 22.643 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 636.306 1 636.306 1.241 Error 23071.662 45 512.704 ------
Test: Test for possible differences in CS freezing based on physical test context
Number of Groups: 2 Number of Measurements: 1 Number of subjects in... Group 1 (Context 1): 26 Group 2 (Context 2): 21
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
369 Means and Standard Deviations Group 1 Overall Mean: 48.722 Measurement 1 Mean 48.722 SD 22.666
Group 2 Overall Mean: 43.493 Measurement 1 Mean 43.493 SD 22.964
Means and SDs averaged across groups Measurement 1 Mean 46.107 SD 22.799 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 317.669 1 317.669 0.611 Error 23390.298 45 519.784 ------
Test: Test for possible differences in CS freezing between groups AAA and ABB
Number of Groups: 2 Number of Measurements: 1 Number of subjects in... Group 1 (AAA): 7 Group 2 (ABB): 7
370 Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 31.671 Measurement 1 Mean 31.671 SD 11.881
Group 2 Overall Mean: 41.671 Measurement 1 Mean 41.671 SD 20.748
Means and SDs averaged across groups Measurement 1 Mean 36.671 SD 16.906 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 350.000 1 350.000 1.225 Error 3429.789 12 285.816 ------
371 Test: Between-groups comparisons of CS freezing
Number of Groups: 3 Number of Measurements: 1
Number of subjects in... Group 1 (Control): 18 Group 2 (Same): 14 Group 3 (Different): 15
Between contrast coefficients Contrast Group... 1 2 3 B1 2 -1 -1 B2 0 1 -1
Means and Standard Deviations Group 1 Overall Mean: 57.806 Measurement 1 Mean 57.806 SD 22.371
Group 2 Overall Mean: 36.671 Measurement 1 Mean 36.671 SD 17.051
Group 3 Overall Mean: 41.793 Measurement 1 Mean 41.793 SD 23.068
372 Means and SDs averaged across groups Measurement 1 Mean 45.423 SD 21.180 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 3829.543 1 3829.543 8.537 B2 189.970 1 189.970 0.423 Error 19737.147 44 448.572 ------
Test: Between-groups comparisons of CS freezing with group AAA removed from group Same
Number of Groups: 3 Number of Measurements: 1
Number of subjects in... Group 1 (Control): 18 Group 2 (ABB): 7 Group 3 (Different): 15
Between contrast coefficients Contrast Group... 1 2 3 B1 2 -1 -1 B2 0 1 -1
373 Means and Standard Deviations Group 1 Overall Mean: 57.806 Measurement 1 Mean 57.806 SD 22.371
Group 2 Overall Mean: 41.671 Measurement 1 Mean 41.671 SD 20.748
Group 3 Overall Mean: 41.793 Measurement 1 Mean 41.793 SD 23.068
Means and SDs averaged across groups Measurement 1 Mean 47.090 SD 22.385 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2393.508 1 2393.508 4.777 B2 0.071 1 0.071 0.000 Error 18540.253 37 501.088 ------
374 Appendix 12: Experiment 10 Raw Data Extinction Day 1 AAA group pre- rat CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 1 90 40 93.3 100 93.3 86.7 80 86.7 86.7 3 96.7 93.3 100 80 93.3 100 93.3 86.7 33.3 4 46.7 26.7 100 46.7 40 40 93.3 80 86.7 7 60 60 93.3 100 100 33.3 100 93.3 100 9 36.7 66.7 73.3 60 40 80 60 33.3 20 mean 66.0 57.392.0 77.3 73.3 68.0 85.3 76.0 65.3 sem 11.8 11.54.9 10.7 13.7 13.2 7.1 10.9 16.1
ABA group pre- rat CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 2 53.3 13.3 66.7 66.7 40 73.3 60 53.3 66.7 5 43.3 53.3 60 73.3 53.3 26.7 60 60 80 6 20 93.3 93.3 93.3 93.3 93.3 73.3 73.3 66.7 8 46.7 6.7 46.7 66.7 40 40 6.7 40 60 10 33.3 73.3 66.7 93.3 93.3 86.7 86.7 73.3 73.3 mean 39.3 48.066.7 78.7 64.0 64.0 57.3 60.0 69.3 sem 5.8 16.87.6 6.1 12.2 13.1 13.6 6.3 3.4
Day 2 AAA group pre- rat CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 1 20 46.7 86.7 0 13.3 66.7 26.7 53.3 60 3 90 73.3 93.3 80 66.7 13.3 86.7 80 13.3 4 10 73.3 20 0 46.7 93.3 0 40 60 7 43.3 86.7 93.3 86.7 20 66.7 80 86.7 86.7 9 13.3 86.7 73.3 26.7 20 26.7 33.3 66.7 33.3 mean 35.3 73.373.3 38.7 33.3 53.3 45.3 65.3 50.7 sem 14.9 7.313.8 18.9 10.1 14.6 16.5 8.5 12.6
ABA group pre- rat CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 2 0 53.3 53.3 53.3 20 13.3 40 6.7 0 5 6.7 46.7 66.7 60 13.3 53.3 46.7 60 80 6 3.3 80 73.3 66.7 66.7 40 53.3 60 46.7 8 0 40 33.3 40 26.7 13.3 33.3 40 13.3 10 3.3 66.7 80 100 13.3 53.3 86.7 86.7 86.7 mean 2.7 57.361.3 64.0 28.0 34.6 52.0 50.7 45.3 sem 1.3 7.28.3 10.0 10.0 9.0 9.3 13.3 17.3
375 Day 3 AAA group pre- rat CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 1 13.3 46.7 13.3 6.7 6.7 26.7 13.3 13.3 0 3 40 26.7 6.7 40 0 6.7 26.7 0 26.7 4 13.3 13.3 53.3 0 73.3 0 13.3 86.7 93.3 7 20 86.7 93.3 60 80 60 53.3 93.3 13.3 9 0 53.3 26.7 0 6.7 0 0 0 0 mean 17.3 45.3 38.7 21.3 33.3 18.7 21.3 38.7 26.7 sem 6.5 12.5 15.8 12.2 17.8 11.4 9.0 21.1 17.4
ABA group pre- rat CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 2 0 46.7 0 6.7 0 20 26.7 0 20 5 10 26.7 40 46.7 0 13.3 0 13.3 13.3 6 13.3 93.3 53.3 40 66.7 6.7 40 46.7 46.7 8 0 0 13.3 0 6.7 33.3 0 13.3 46.7 10 0 86.7 13.3 0 86.7 46.7 0 0 0 mean 4.7 50.7 24.0 18.7 32.0 24.0 13.3 14.7 25.3 sem 2.9 17.7 9.8 10.218.6 7.2 8.4 8.5 9.3
Test AAA group ABA group pre- pre- rat CS CS1 rat CS CS1 1 0 20 2 3.3 40 3 13.3 6.7 5 6.7 46.7 4 0 6.7 6 46.7 73.3 7 10 46.7 8 10 46.7 9 0 0 10 26.7 86.7 mean 4.7 16.0 mean 18.7 58.7 sem 2.9 8.3 sem 8.1 9.0
Statistical Analyses
Extinction: Between-groups analysis of context freezing
Number of Groups: 2 Number of Measurements: 3
Number of subjects in... Group 1 (AAA): 5 Group 2 (ABA): 5
376 Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 W1 -1 0 1
Means and Standard Deviations Group 1 Overall Mean: 39.553 Measurement 1 2 3 Mean 66.020 35.320 17.320 SD 26.389 33.220 14.611
Group 2 Overall Mean: 15.547 Measurement 1 2 3 Mean 39.320 2.660 4.660 SD 12.993 2.797 6.487
Means and SDs averaged across groups Measurement 1 2 3 Mean 52.670 18.990 10.990 SD 20.799 23.573 11.304 ------
377 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 4322.400 1 4322.400 5.397 Error 6407.381 8 800.923 ------Within ------W1 8686.112 1 8686.112 53.875 B1W1 246.402 1 246.402 1.528 Error 1289.826 8 161.228 ------
Extinction: Between-groups analysis of CS freezing
Number of Groups: 2 Number of Measurements: 24
Number of subjects in... Group 1 (AAA): 5 Group 2 (ABA): 5
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 W1 -23 -21 -19 -17 -15 -13 -11 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 21 23
378 Means and Standard Deviations Group 1 Overall Mean: 53.000 Measurement 1 2 3 4 5 6 7 8 Mean 57.340 91.980 77.340 73.320 68.000 85.320 76.000 65.340 SD 25.629 10.967 23.841 30.540 29.605 15.908 24.329 36.042 9 10 11 12 13 14 15 16 17 18 73.340 73.320 38.680 33.340 53.340 45.340 65.340 50.660 45.340 38.660 16.330 30.905 42.276 22.632 32.660 36.946 19.106 28.154 28.062 35.382 19 20 21 22 23 24 21.340 33.340 18.680 21.320 38.660 26.660 27.239 39.702 25.559 20.217 47.238 38.862
Group 2 Overall Mean: 45.999 Measurement 1 2 3 4 5 6 7 8 Mean 47.980 66.680 78.660 63.980 64.000 57.340 59.980 69.340 SD 37.517 16.974 13.633 27.311 29.273 30.394 14.130 7.591 9 10 11 12 13 14 15 16 17 18 57.340 61.320 64.000 28.000 34.640 52.000 50.680 45.340 50.680 23.980 16.057 18.512 22.415 22.336 20.223 20.782 29.663 38.717 39.602 21.906 19 20 21 22 23 24 18.680 32.020 24.000 13.340 14.660 25.340 22.809 41.486 16.059 18.862 19.106 20.785
Means and SDs averaged across groups Measurement 1 2 3 4 5 6 7 8 Mean 52.660 79.330 78.000 68.650 66.000 71.330 67.990 67.340 SD 32.127 14.289 19.420 28.970 29.439 24.258 19.894 26.045 9 10 11 12 13 14 15 16 17 18 65.340 67.320 51.340 30.670 43.990 48.670 58.010 48.000 48.010 31.320 16.194 25.474 33.836 22.485 27.163 29.974 24.949 33.850 34.320 29.426 19 20 21 22 23 24 20.010 32.680 21.340 17.330 26.660 26.000 25.122 40.604 21.344 19.551 36.031 31.163
379 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 2940.700 1 2940.700 0.511 Error 46052.943 8 5756.618 ------Within ------W1 67764.755 1 67764.755 39.942 B1W1 246.172 1 246.172 0.145 Error 13572.757 8 1696.595 ------
Extinction: Analysis of effects of physical context on context freezing
Number of Groups: 2 Number of Measurements: 3
Number of subjects in... Group 1 (context 2): 6 Group 2 (context 1): 4
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients Contrast Measurement... 1 2 3 W1 -1 0 1
380 Means and Standard Deviations Group 1 Overall Mean: 33.883 Measurement 1 2 3 Mean 61.117 26.100 14.433 SD 28.500 35.492 14.858
Group 2 Overall Mean: 18.050 Measurement 1 2 3 Mean 40.000 8.325 5.825 SD 6.098 4.299 6.860
Means and SDs averaged across groups Measurement 1 2 3 Mean 50.558 17.213 10.129 SD 22.838 28.182 12.475 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 1805.000 1 1805.000 1.618 Error 8924.782 8 1115.598 ------Within ------W1 7845.684 1 7845.684 46.545 B1W1 187.750 1 187.750 1.114 Error 1348.478 8 168.560 ------
381 Extinction: Analysis of effects of physical context on CS freezing
Number of Groups: 2 Number of Measurements: 24
Number of subjects in... Group 1 (context 1): 4 Group 2 (context 2): 6
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Within contrast coefficients 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 W1 -23 -21 -19 -17 -15 -13 -11 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 21 23
Means and Standard Deviations Group 1 Overall Mean: 47.569 Measurement 1 2 3 4 5 6 7 8 Mean 55.000 75.000 68.325 56.650 58.350 75.000 61.650 65.000 SD 20.620 17.529 19.878 25.225 29.501 17.529 20.647 30.495 9 10 11 12 13 14 15 16 17 18 68.350 60.000 46.675 23.325 56.650 41.675 63.350 65.000 45.000 33.325 16.660 27.214 43.200 15.900 27.463 35.868 19.260 23.985 32.391 17.209 19 20 21 22 23 24 11.675 41.675 15.000 3.325 25.000 26.650 23.350 44.675 22.044 6.650 41.608 44.873
382 Group 2 Overall Mean: 50.787 Measurement 1 2 3 4 5 6 7 8 Mean 51.100 82.217 84.450 76.650 71.100 68.883 72.217 68.900 SD 37.853 20.914 15.568 28.507 28.186 33.625 21.265 22.980 9 10 11 12 13 14 15 16 17 18 63.333 72.200 54.450 35.567 35.550 53.333 54.450 36.667 50.017 29.983 19.209 24.382 31.679 24.485 26.251 24.952 29.031 33.421 35.458 36.195 19 20 21 22 23 24 25.567 26.683 25.567 26.667 27.767 25.567 24.363 36.515 19.948 18.851 36.369 18.599
Means and SDs averaged across groups Measurement 1 2 3 4 5 6 7 8 Mean 53.050 78.608 76.388 66.650 64.725 71.942 66.933 66.950 SD 32.480 19.713 17.311 27.323 28.686 28.668 21.035 26.053 9 10 11 12 13 14 15 16 17 18 65.842 66.100 50.563 29.446 46.100 47.504 58.900 50.833 47.508 31.654 18.295 25.481 36.429 21.668 26.712 29.522 25.804 30.230 34.340 30.494 19 20 21 22 23 24 18.621 34.179 20.283 14.996 26.383 26.108 23.988 39.771 20.759 15.450 38.418 31.166 ------
383 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 596.499 1 596.499 0.099 Error 48397.144 8 6049.643 ------Within ------W1 63568.322 1 63568.322 37.381 B1W1 214.562 1 214.562 0.126 Error 13604.366 8 1700.546 ------
Test: Between-group contrast of context freezing
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (AAA): 5 Group 2 (ABA): 5
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 4.660 Measurement 1 Mean 4.660 SD 6.487
384 Group 2 Overall Mean: 18.680 Measurement 1 Mean 18.680 SD 18.061
Means and SDs averaged across groups Measurement 1 Mean 11.670 SD 13.570 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 491.401 1 491.401 2.669 Error 1473.160 8 184.145 ------
Test: Between-groups contrast of CS freezing.
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (AAA): 5 Group 2 (ABA): 5
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
385 Means and Standard Deviations Group 1 Overall Mean: 16.020 Measurement 1 Mean 16.020 SD 18.624
Group 2 Overall Mean: 58.680 Measurement 1 Mean 58.680 SD 20.217
Means and SDs averaged across groups Measurement 1 Mean 37.350 SD 19.437 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 4549.689 1 4549.689 12.043 Error 3022.316 8 377.790 ------
Test: Effect of physical context on context freezing.
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (context 1): 4 Group 2 (context 2): 6
386 Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 8.350 Measurement 1 Mean 8.350 SD 12.634
Group 2 Overall Mean: 13.883 Measurement 1 Mean 13.883 SD 16.806
Means and SDs averaged across groups Measurement 1 Mean 11.117 SD 15.375 ------
Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 73.483 1 73.483 0.311 Error 1891.078 8 236.385 ------
387 Test: Effect of physical context on CS freezing.
Number of Groups: 2 Number of Measurements: 1
Number of subjects in... Group 1 (context 1): 4 Group 2 (context 2): 6
Between contrast coefficients Contrast Group... 1 2 B1 1 -1
Means and Standard Deviations Group 1 Overall Mean: 35.025 Measurement 1 Mean 35.025 SD 40.148
Group 2 Overall Mean: 38.900 Measurement 1 Mean 38.900 SD 23.239
Means and SDs averaged across groups Measurement 1 Mean 36.963 SD 30.692 ------
388 Analysis of Variance Summary Table Source SS df MS F ------Between ------B1 36.037 1 36.037 0.038 Error 7535.968 8 941.996 ------
389